Merge git://git.kernel.org/pub/scm/linux/kernel/git/lethal/sh-2.6

* git://git.kernel.org/pub/scm/linux/kernel/git/lethal/sh-2.6: (26 commits)
  sh: Convert sh to use read/update_persistent_clock
  sh: Move PMB debugfs entry initialization to later stage
  sh: Fix up flush_cache_vmap() on SMP.
  sh: fix up MMU reset with variable PMB mapping sizes.
  sh: establish PMB mappings for NUMA nodes.
  sh: check for existing mappings for bolted PMB entries.
  sh: fixed virt/phys mapping helpers for PMB.
  sh: make pmb iomapping configurable.
  sh: reworked dynamic PMB mapping.
  sh: Fix up cpumask_of_pcibus() for the NUMA build.
  serial: sh-sci: Tidy up build warnings.
  sh: Fix up ctrl_read/write stragglers in migor setup.
  serial: sh-sci: Add DMA support.
  dmaengine: shdma: extend .device_terminate_all() to record partial transfer
  sh: merge sh7722 and sh7724 DMA register definitions
  sh: activate runtime PM for dmaengine on sh7722 and sh7724
  dmaengine: shdma: add runtime PM support.
  dmaengine: shdma: separate DMA headers.
  dmaengine: shdma: convert to platform device resources
  dmaengine: shdma: fix DMA error handling.
  ...

35 files changed, 2201 insertions, 728 deletions

diff --git a/arch/sh/boards/mach-migor/setup.c b/arch/sh/boards/mach-migor/setup.c
index be300aaca6fe..7da0fc94a01e 100644
--- a/arch/sh/boards/mach-migor/setup.c
+++ b/arch/sh/boards/mach-migor/setup.c
@@ -419,6 +419,9 @@ static struct i2c_board_info migor_i2c_devices[] = {
 		I2C_BOARD_INFO("migor_ts", 0x51),
 		.irq = 38, /* IRQ6 */
 	},
+	{
+		I2C_BOARD_INFO("wm8978", 0x1a),
+	},
 };
 
 static struct i2c_board_info migor_i2c_camera[] = {
@@ -619,6 +622,19 @@ static int __init migor_devices_setup(void)
 
 	platform_resource_setup_memory(&migor_ceu_device, "ceu", 4 << 20);
 
+	/* SIU: Port B */
+	gpio_request(GPIO_FN_SIUBOLR, NULL);
+	gpio_request(GPIO_FN_SIUBOBT, NULL);
+	gpio_request(GPIO_FN_SIUBISLD, NULL);
+	gpio_request(GPIO_FN_SIUBOSLD, NULL);
+	gpio_request(GPIO_FN_SIUMCKB, NULL);
+
+	/*
+	 * The original driver sets SIUB OLR/OBT, ILR/IBT, and SIUA OLR/OBT to
+	 * output. Need only SIUB, set to output for master mode (table 34.2)
+	 */
+	__raw_writew(__raw_readw(PORT_MSELCRA) | 1, PORT_MSELCRA);
+
 	i2c_register_board_info(0, migor_i2c_devices,
 				ARRAY_SIZE(migor_i2c_devices));
 
diff --git a/arch/sh/boot/compressed/cache.c b/arch/sh/boot/compressed/cache.c
index e27fc74f228c..d0b77b68a4d0 100644
--- a/arch/sh/boot/compressed/cache.c
+++ b/arch/sh/boot/compressed/cache.c
@@ -5,7 +5,7 @@ int cache_control(unsigned int command)
 
 	for (i = 0; i < (32 * 1024); i += 32) {
 		(void)*p;
-		p += (32 / sizeof (int));
+		p += (32 / sizeof(int));
 	}
 
 	return 0;
diff --git a/arch/sh/include/asm/cacheflush.h b/arch/sh/include/asm/cacheflush.h
index da3ebec921a7..1f4e562c5e8c 100644
--- a/arch/sh/include/asm/cacheflush.h
+++ b/arch/sh/include/asm/cacheflush.h
@@ -86,8 +86,8 @@ extern void copy_from_user_page(struct vm_area_struct *vma,
 	struct page *page, unsigned long vaddr, void *dst, const void *src,
 	unsigned long len);
 
-#define flush_cache_vmap(start, end)		flush_cache_all()
-#define flush_cache_vunmap(start, end)		flush_cache_all()
+#define flush_cache_vmap(start, end)		local_flush_cache_all(NULL)
+#define flush_cache_vunmap(start, end)		local_flush_cache_all(NULL)
 
 #define flush_dcache_mmap_lock(mapping)		do { } while (0)
 #define flush_dcache_mmap_unlock(mapping)	do { } while (0)
diff --git a/arch/sh/include/asm/dma-register.h b/arch/sh/include/asm/dma-register.h
new file mode 100644
index 000000000000..51cd78feacff
--- /dev/null
+++ b/arch/sh/include/asm/dma-register.h
@@ -0,0 +1,51 @@
+/*
+ * Common header for the legacy SH DMA driver and the new dmaengine driver
+ *
+ * extracted from arch/sh/include/asm/dma-sh.h:
+ *
+ * Copyright (C) 2000  Takashi YOSHII
+ * Copyright (C) 2003  Paul Mundt
+ *
+ * This file is subject to the terms and conditions of the GNU General Public
+ * License.  See the file "COPYING" in the main directory of this archive
+ * for more details.
+ */
+#ifndef DMA_REGISTER_H
+#define DMA_REGISTER_H
+
+/* DMA register */
+#define SAR	0x00
+#define DAR	0x04
+#define TCR	0x08
+#define CHCR	0x0C
+#define DMAOR	0x40
+
+/* DMAOR definitions */
+#define DMAOR_AE	0x00000004
+#define DMAOR_NMIF	0x00000002
+#define DMAOR_DME	0x00000001
+
+/* Definitions for the SuperH DMAC */
+#define REQ_L	0x00000000
+#define REQ_E	0x00080000
+#define RACK_H	0x00000000
+#define RACK_L	0x00040000
+#define ACK_R	0x00000000
+#define ACK_W	0x00020000
+#define ACK_H	0x00000000
+#define ACK_L	0x00010000
+#define DM_INC	0x00004000
+#define DM_DEC	0x00008000
+#define DM_FIX	0x0000c000
+#define SM_INC	0x00001000
+#define SM_DEC	0x00002000
+#define SM_FIX	0x00003000
+#define RS_IN	0x00000200
+#define RS_OUT	0x00000300
+#define TS_BLK	0x00000040
+#define TM_BUR	0x00000020
+#define CHCR_DE	0x00000001
+#define CHCR_TE	0x00000002
+#define CHCR_IE	0x00000004
+
+#endif
diff --git a/arch/sh/include/asm/dma-sh.h b/arch/sh/include/asm/dma-sh.h
index e934a2e66651..f3acb8e34c6b 100644
--- a/arch/sh/include/asm/dma-sh.h
+++ b/arch/sh/include/asm/dma-sh.h
@@ -11,7 +11,8 @@
 #ifndef __DMA_SH_H
 #define __DMA_SH_H
 
-#include <asm/dma.h>
+#include <asm/dma-register.h>
+#include <cpu/dma-register.h>
 #include <cpu/dma.h>
 
 /* DMAOR contorl: The DMAOR access size is different by CPU.*/
@@ -53,34 +54,6 @@ static int dmte_irq_map[] __maybe_unused = {
 #endif
 };
 
-/* Definitions for the SuperH DMAC */
-#define REQ_L	0x00000000
-#define REQ_E	0x00080000
-#define RACK_H	0x00000000
-#define RACK_L	0x00040000
-#define ACK_R	0x00000000
-#define ACK_W	0x00020000
-#define ACK_H	0x00000000
-#define ACK_L	0x00010000
-#define DM_INC	0x00004000
-#define DM_DEC	0x00008000
-#define DM_FIX	0x0000c000
-#define SM_INC	0x00001000
-#define SM_DEC	0x00002000
-#define SM_FIX	0x00003000
-#define RS_IN	0x00000200
-#define RS_OUT	0x00000300
-#define TS_BLK	0x00000040
-#define TM_BUR	0x00000020
-#define CHCR_DE 0x00000001
-#define CHCR_TE 0x00000002
-#define CHCR_IE 0x00000004
-
-/* DMAOR definitions */
-#define DMAOR_AE	0x00000004
-#define DMAOR_NMIF	0x00000002
-#define DMAOR_DME	0x00000001
-
 /*
  * Define the default configuration for dual address memory-memory transfer.
  * The 0x400 value represents auto-request, external->external.
@@ -111,61 +84,4 @@ static u32 dma_base_addr[] __maybe_unused = {
 #endif
 };
 
-/* DMA register */
-#define SAR     0x00
-#define DAR     0x04
-#define TCR     0x08
-#define CHCR    0x0C
-#define DMAOR	0x40
-
-/*
- * for dma engine
- *
- * SuperH DMA mode
- */
-#define SHDMA_MIX_IRQ	(1 << 1)
-#define SHDMA_DMAOR1	(1 << 2)
-#define SHDMA_DMAE1	(1 << 3)
-
-enum sh_dmae_slave_chan_id {
-	SHDMA_SLAVE_SCIF0_TX,
-	SHDMA_SLAVE_SCIF0_RX,
-	SHDMA_SLAVE_SCIF1_TX,
-	SHDMA_SLAVE_SCIF1_RX,
-	SHDMA_SLAVE_SCIF2_TX,
-	SHDMA_SLAVE_SCIF2_RX,
-	SHDMA_SLAVE_SCIF3_TX,
-	SHDMA_SLAVE_SCIF3_RX,
-	SHDMA_SLAVE_SCIF4_TX,
-	SHDMA_SLAVE_SCIF4_RX,
-	SHDMA_SLAVE_SCIF5_TX,
-	SHDMA_SLAVE_SCIF5_RX,
-	SHDMA_SLAVE_SIUA_TX,
-	SHDMA_SLAVE_SIUA_RX,
-	SHDMA_SLAVE_SIUB_TX,
-	SHDMA_SLAVE_SIUB_RX,
-	SHDMA_SLAVE_NUMBER,	/* Must stay last */
-};
-
-struct sh_dmae_slave_config {
-	enum sh_dmae_slave_chan_id	slave_id;
-	dma_addr_t			addr;
-	u32				chcr;
-	char				mid_rid;
-};
-
-struct sh_dmae_pdata {
-	unsigned int mode;
-	struct sh_dmae_slave_config *config;
-	int config_num;
-};
-
-struct device;
-
-struct sh_dmae_slave {
-	enum sh_dmae_slave_chan_id	slave_id; /* Set by the platform */
-	struct device			*dma_dev; /* Set by the platform */
-	struct sh_dmae_slave_config	*config;  /* Set by the driver */
-};
-
 #endif /* __DMA_SH_H */
diff --git a/arch/sh/include/asm/dmaengine.h b/arch/sh/include/asm/dmaengine.h
new file mode 100644
index 000000000000..bf2f30cf0a27
--- /dev/null
+++ b/arch/sh/include/asm/dmaengine.h
@@ -0,0 +1,93 @@
+/*
+ * Header for the new SH dmaengine driver
+ *
+ * Copyright (C) 2010 Guennadi Liakhovetski <g.liakhovetski@gmx.de>
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ */
+#ifndef ASM_DMAENGINE_H
+#define ASM_DMAENGINE_H
+
+#include <linux/dmaengine.h>
+#include <linux/list.h>
+
+#include <asm/dma-register.h>
+
+#define SH_DMAC_MAX_CHANNELS	6
+
+enum sh_dmae_slave_chan_id {
+	SHDMA_SLAVE_SCIF0_TX,
+	SHDMA_SLAVE_SCIF0_RX,
+	SHDMA_SLAVE_SCIF1_TX,
+	SHDMA_SLAVE_SCIF1_RX,
+	SHDMA_SLAVE_SCIF2_TX,
+	SHDMA_SLAVE_SCIF2_RX,
+	SHDMA_SLAVE_SCIF3_TX,
+	SHDMA_SLAVE_SCIF3_RX,
+	SHDMA_SLAVE_SCIF4_TX,
+	SHDMA_SLAVE_SCIF4_RX,
+	SHDMA_SLAVE_SCIF5_TX,
+	SHDMA_SLAVE_SCIF5_RX,
+	SHDMA_SLAVE_SIUA_TX,
+	SHDMA_SLAVE_SIUA_RX,
+	SHDMA_SLAVE_SIUB_TX,
+	SHDMA_SLAVE_SIUB_RX,
+	SHDMA_SLAVE_NUMBER,	/* Must stay last */
+};
+
+struct sh_dmae_slave_config {
+	enum sh_dmae_slave_chan_id	slave_id;
+	dma_addr_t			addr;
+	u32				chcr;
+	char				mid_rid;
+};
+
+struct sh_dmae_channel {
+	unsigned int	offset;
+	unsigned int	dmars;
+	unsigned int	dmars_bit;
+};
+
+struct sh_dmae_pdata {
+	struct sh_dmae_slave_config *slave;
+	int slave_num;
+	struct sh_dmae_channel *channel;
+	int channel_num;
+	unsigned int ts_low_shift;
+	unsigned int ts_low_mask;
+	unsigned int ts_high_shift;
+	unsigned int ts_high_mask;
+	unsigned int *ts_shift;
+	int ts_shift_num;
+	u16 dmaor_init;
+};
+
+struct device;
+
+/* Used by slave DMA clients to request DMA to/from a specific peripheral */
+struct sh_dmae_slave {
+	enum sh_dmae_slave_chan_id	slave_id; /* Set by the platform */
+	struct device			*dma_dev; /* Set by the platform */
+	struct sh_dmae_slave_config	*config;  /* Set by the driver */
+};
+
+struct sh_dmae_regs {
+	u32 sar; /* SAR / source address */
+	u32 dar; /* DAR / destination address */
+	u32 tcr; /* TCR / transfer count */
+};
+
+struct sh_desc {
+	struct sh_dmae_regs hw;
+	struct list_head node;
+	struct dma_async_tx_descriptor async_tx;
+	enum dma_data_direction direction;
+	dma_cookie_t cookie;
+	size_t partial;
+	int chunks;
+	int mark;
+};
+
+#endif
diff --git a/arch/sh/include/asm/io.h b/arch/sh/include/asm/io.h
index 7dab7b23a5ec..f689554e17c1 100644
--- a/arch/sh/include/asm/io.h
+++ b/arch/sh/include/asm/io.h
@@ -291,21 +291,21 @@ unsigned long long poke_real_address_q(unsigned long long addr,
  * doesn't exist, so everything must go through page tables.
  */
 #ifdef CONFIG_MMU
-void __iomem *__ioremap_caller(unsigned long offset, unsigned long size,
+void __iomem *__ioremap_caller(phys_addr_t offset, unsigned long size,
 			       pgprot_t prot, void *caller);
 void __iounmap(void __iomem *addr);
 
 static inline void __iomem *
-__ioremap(unsigned long offset, unsigned long size, pgprot_t prot)
+__ioremap(phys_addr_t offset, unsigned long size, pgprot_t prot)
 {
 	return __ioremap_caller(offset, size, prot, __builtin_return_address(0));
 }
 
 static inline void __iomem *
-__ioremap_29bit(unsigned long offset, unsigned long size, pgprot_t prot)
+__ioremap_29bit(phys_addr_t offset, unsigned long size, pgprot_t prot)
 {
 #ifdef CONFIG_29BIT
-	unsigned long last_addr = offset + size - 1;
+	phys_addr_t last_addr = offset + size - 1;
 
 	/*
 	 * For P1 and P2 space this is trivial, as everything is already
@@ -329,7 +329,7 @@ __ioremap_29bit(unsigned long offset, unsigned long size, pgprot_t prot)
 }
 
 static inline void __iomem *
-__ioremap_mode(unsigned long offset, unsigned long size, pgprot_t prot)
+__ioremap_mode(phys_addr_t offset, unsigned long size, pgprot_t prot)
 {
 	void __iomem *ret;
 
@@ -349,35 +349,32 @@ __ioremap_mode(unsigned long offset, unsigned long size, pgprot_t prot)
 #define __iounmap(addr)				do { } while (0)
 #endif /* CONFIG_MMU */
 
-static inline void __iomem *
-ioremap(unsigned long offset, unsigned long size)
+static inline void __iomem *ioremap(phys_addr_t offset, unsigned long size)
 {
 	return __ioremap_mode(offset, size, PAGE_KERNEL_NOCACHE);
 }
 
 static inline void __iomem *
-ioremap_cache(unsigned long offset, unsigned long size)
+ioremap_cache(phys_addr_t offset, unsigned long size)
 {
 	return __ioremap_mode(offset, size, PAGE_KERNEL);
 }
 
 #ifdef CONFIG_HAVE_IOREMAP_PROT
 static inline void __iomem *
-ioremap_prot(resource_size_t offset, unsigned long size, unsigned long flags)
+ioremap_prot(phys_addr_t offset, unsigned long size, unsigned long flags)
 {
 	return __ioremap_mode(offset, size, __pgprot(flags));
 }
 #endif
 
 #ifdef CONFIG_IOREMAP_FIXED
-extern void __iomem *ioremap_fixed(resource_size_t, unsigned long,
-				   unsigned long, pgprot_t);
+extern void __iomem *ioremap_fixed(phys_addr_t, unsigned long, pgprot_t);
 extern int iounmap_fixed(void __iomem *);
 extern void ioremap_fixed_init(void);
 #else
 static inline void __iomem *
-ioremap_fixed(resource_size_t phys_addr, unsigned long offset,
-	      unsigned long size, pgprot_t prot)
+ioremap_fixed(phys_addr_t phys_addr, unsigned long size, pgprot_t prot)
 {
 	BUG();
 	return NULL;
diff --git a/arch/sh/include/asm/mmu.h b/arch/sh/include/asm/mmu.h
index 15a05b615ba7..19fe84550b49 100644
--- a/arch/sh/include/asm/mmu.h
+++ b/arch/sh/include/asm/mmu.h
@@ -55,19 +55,29 @@ typedef struct {
 
 #ifdef CONFIG_PMB
 /* arch/sh/mm/pmb.c */
-long pmb_remap(unsigned long virt, unsigned long phys,
-	       unsigned long size, pgprot_t prot);
-void pmb_unmap(unsigned long addr);
-void pmb_init(void);
 bool __in_29bit_mode(void);
+
+void pmb_init(void);
+int pmb_bolt_mapping(unsigned long virt, phys_addr_t phys,
+		     unsigned long size, pgprot_t prot);
+void __iomem *pmb_remap_caller(phys_addr_t phys, unsigned long size,
+			       pgprot_t prot, void *caller);
+int pmb_unmap(void __iomem *addr);
+
 #else
-static inline long pmb_remap(unsigned long virt, unsigned long phys,
-			     unsigned long size, pgprot_t prot)
+
+static inline void __iomem *
+pmb_remap_caller(phys_addr_t phys, unsigned long size,
+		 pgprot_t prot, void *caller)
+{
+	return NULL;
+}
+
+static inline int pmb_unmap(void __iomem *addr)
 {
 	return -EINVAL;
 }
 
-#define pmb_unmap(addr)		do { } while (0)
 #define pmb_init(addr)		do { } while (0)
 
 #ifdef CONFIG_29BIT
@@ -77,6 +87,13 @@ static inline long pmb_remap(unsigned long virt, unsigned long phys,
 #endif
 
 #endif /* CONFIG_PMB */
+
+static inline void __iomem *
+pmb_remap(phys_addr_t phys, unsigned long size, pgprot_t prot)
+{
+	return pmb_remap_caller(phys, size, prot, __builtin_return_address(0));
+}
+
 #endif /* __ASSEMBLY__ */
 
 #endif /* __MMU_H */
diff --git a/arch/sh/include/asm/siu.h b/arch/sh/include/asm/siu.h
index 57565a3b551f..f1b1e6944a5f 100644
--- a/arch/sh/include/asm/siu.h
+++ b/arch/sh/include/asm/siu.h
@@ -11,7 +11,7 @@
 #ifndef ASM_SIU_H
 #define ASM_SIU_H
 
-#include <asm/dma-sh.h>
+#include <asm/dmaengine.h>
 
 struct device;
 
diff --git a/arch/sh/include/asm/topology.h b/arch/sh/include/asm/topology.h
index 37cdadd975ac..88e734069fa6 100644
--- a/arch/sh/include/asm/topology.h
+++ b/arch/sh/include/asm/topology.h
@@ -35,7 +35,7 @@
 
 #define pcibus_to_node(bus)	((void)(bus), -1)
 #define cpumask_of_pcibus(bus)	(pcibus_to_node(bus) == -1 ? \
-					CPU_MASK_ALL_PTR : \
+					cpu_all_mask : \
 					cpumask_of_node(pcibus_to_node(bus)))
 
 #endif
diff --git a/arch/sh/include/cpu-sh3/cpu/dma-register.h b/arch/sh/include/cpu-sh3/cpu/dma-register.h
new file mode 100644
index 000000000000..2349e488c9a6
--- /dev/null
+++ b/arch/sh/include/cpu-sh3/cpu/dma-register.h
@@ -0,0 +1,41 @@
+/*
+ * SH3 CPU-specific DMA definitions, used by both DMA drivers
+ *
+ * Copyright (C) 2010 Guennadi Liakhovetski <g.liakhovetski@gmx.de>
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ */
+#ifndef CPU_DMA_REGISTER_H
+#define CPU_DMA_REGISTER_H
+
+#define CHCR_TS_LOW_MASK	0x18
+#define CHCR_TS_LOW_SHIFT	3
+#define CHCR_TS_HIGH_MASK	0
+#define CHCR_TS_HIGH_SHIFT	0
+
+#define DMAOR_INIT	DMAOR_DME
+
+/*
+ * The SuperH DMAC supports a number of transmit sizes, we list them here,
+ * with their respective values as they appear in the CHCR registers.
+ */
+enum {
+	XMIT_SZ_8BIT,
+	XMIT_SZ_16BIT,
+	XMIT_SZ_32BIT,
+	XMIT_SZ_128BIT,
+};
+
+/* log2(size / 8) - used to calculate number of transfers */
+#define TS_SHIFT {			\
+	[XMIT_SZ_8BIT]		= 0,	\
+	[XMIT_SZ_16BIT]		= 1,	\
+	[XMIT_SZ_32BIT]		= 2,	\
+	[XMIT_SZ_128BIT]	= 4,	\
+}
+
+#define TS_INDEX2VAL(i)	(((i) & 3) << CHCR_TS_LOW_SHIFT)
+
+#endif
diff --git a/arch/sh/include/cpu-sh3/cpu/dma.h b/arch/sh/include/cpu-sh3/cpu/dma.h
index 207811a7a650..24e28b91c9d5 100644
--- a/arch/sh/include/cpu-sh3/cpu/dma.h
+++ b/arch/sh/include/cpu-sh3/cpu/dma.h
@@ -20,31 +20,4 @@
 #define TS_32		0x00000010
 #define TS_128		0x00000018
 
-#define CHCR_TS_LOW_MASK	0x18
-#define CHCR_TS_LOW_SHIFT	3
-#define CHCR_TS_HIGH_MASK	0
-#define CHCR_TS_HIGH_SHIFT	0
-
-#define DMAOR_INIT	DMAOR_DME
-
-/*
- * The SuperH DMAC supports a number of transmit sizes, we list them here,
- * with their respective values as they appear in the CHCR registers.
- */
-enum {
-	XMIT_SZ_8BIT,
-	XMIT_SZ_16BIT,
-	XMIT_SZ_32BIT,
-	XMIT_SZ_128BIT,
-};
-
-#define TS_SHIFT {			\
-	[XMIT_SZ_8BIT]		= 0,	\
-	[XMIT_SZ_16BIT]		= 1,	\
-	[XMIT_SZ_32BIT]		= 2,	\
-	[XMIT_SZ_128BIT]	= 4,	\
-}
-
-#define TS_INDEX2VAL(i)	(((i) & 3) << CHCR_TS_LOW_SHIFT)
-
 #endif /* __ASM_CPU_SH3_DMA_H */
diff --git a/arch/sh/include/cpu-sh4/cpu/dma-register.h b/arch/sh/include/cpu-sh4/cpu/dma-register.h
new file mode 100644
index 000000000000..55f9fec082d4
--- /dev/null
+++ b/arch/sh/include/cpu-sh4/cpu/dma-register.h
@@ -0,0 +1,112 @@
+/*
+ * SH4 CPU-specific DMA definitions, used by both DMA drivers
+ *
+ * Copyright (C) 2010 Guennadi Liakhovetski <g.liakhovetski@gmx.de>
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ */
+#ifndef CPU_DMA_REGISTER_H
+#define CPU_DMA_REGISTER_H
+
+/* SH7751/7760/7780 DMA IRQ sources */
+
+#ifdef CONFIG_CPU_SH4A
+
+#define DMAOR_INIT	DMAOR_DME
+
+#if defined(CONFIG_CPU_SUBTYPE_SH7343) || \
+	defined(CONFIG_CPU_SUBTYPE_SH7730)
+#define CHCR_TS_LOW_MASK	0x00000018
+#define CHCR_TS_LOW_SHIFT	3
+#define CHCR_TS_HIGH_MASK	0
+#define CHCR_TS_HIGH_SHIFT	0
+#elif defined(CONFIG_CPU_SUBTYPE_SH7722) || \
+	defined(CONFIG_CPU_SUBTYPE_SH7724)
+#define CHCR_TS_LOW_MASK	0x00000018
+#define CHCR_TS_LOW_SHIFT	3
+#define CHCR_TS_HIGH_MASK	0x00300000
+#define CHCR_TS_HIGH_SHIFT	(20 - 2)	/* 2 bits for shifted low TS */
+#elif defined(CONFIG_CPU_SUBTYPE_SH7763) || \
+	defined(CONFIG_CPU_SUBTYPE_SH7764)
+#define CHCR_TS_LOW_MASK	0x00000018
+#define CHCR_TS_LOW_SHIFT	3
+#define CHCR_TS_HIGH_MASK	0
+#define CHCR_TS_HIGH_SHIFT	0
+#elif defined(CONFIG_CPU_SUBTYPE_SH7723)
+#define CHCR_TS_LOW_MASK	0x00000018
+#define CHCR_TS_LOW_SHIFT	3
+#define CHCR_TS_HIGH_MASK	0
+#define CHCR_TS_HIGH_SHIFT	0
+#elif defined(CONFIG_CPU_SUBTYPE_SH7780)
+#define CHCR_TS_LOW_MASK	0x00000018
+#define CHCR_TS_LOW_SHIFT	3
+#define CHCR_TS_HIGH_MASK	0
+#define CHCR_TS_HIGH_SHIFT	0
+#else /* SH7785 */
+#define CHCR_TS_LOW_MASK	0x00000018
+#define CHCR_TS_LOW_SHIFT	3
+#define CHCR_TS_HIGH_MASK	0
+#define CHCR_TS_HIGH_SHIFT	0
+#endif
+
+/* Transmit sizes and respective CHCR register values */
+enum {
+	XMIT_SZ_8BIT		= 0,
+	XMIT_SZ_16BIT		= 1,
+	XMIT_SZ_32BIT		= 2,
+	XMIT_SZ_64BIT		= 7,
+	XMIT_SZ_128BIT		= 3,
+	XMIT_SZ_256BIT		= 4,
+	XMIT_SZ_128BIT_BLK	= 0xb,
+	XMIT_SZ_256BIT_BLK	= 0xc,
+};
+
+/* log2(size / 8) - used to calculate number of transfers */
+#define TS_SHIFT {			\
+	[XMIT_SZ_8BIT]		= 0,	\
+	[XMIT_SZ_16BIT]		= 1,	\
+	[XMIT_SZ_32BIT]		= 2,	\
+	[XMIT_SZ_64BIT]		= 3,	\
+	[XMIT_SZ_128BIT]	= 4,	\
+	[XMIT_SZ_256BIT]	= 5,	\
+	[XMIT_SZ_128BIT_BLK]	= 4,	\
+	[XMIT_SZ_256BIT_BLK]	= 5,	\
+}
+
+#define TS_INDEX2VAL(i)	((((i) & 3) << CHCR_TS_LOW_SHIFT) | \
+			 ((((i) >> 2) & 3) << CHCR_TS_HIGH_SHIFT))
+
+#else /* CONFIG_CPU_SH4A */
+
+#define DMAOR_INIT	(0x8000 | DMAOR_DME)
+
+#define CHCR_TS_LOW_MASK	0x70
+#define CHCR_TS_LOW_SHIFT	4
+#define CHCR_TS_HIGH_MASK	0
+#define CHCR_TS_HIGH_SHIFT	0
+
+/* Transmit sizes and respective CHCR register values */
+enum {
+	XMIT_SZ_8BIT	= 1,
+	XMIT_SZ_16BIT	= 2,
+	XMIT_SZ_32BIT	= 3,
+	XMIT_SZ_64BIT	= 0,
+	XMIT_SZ_256BIT	= 4,
+};
+
+/* log2(size / 8) - used to calculate number of transfers */
+#define TS_SHIFT {			\
+	[XMIT_SZ_8BIT]		= 0,	\
+	[XMIT_SZ_16BIT]		= 1,	\
+	[XMIT_SZ_32BIT]		= 2,	\
+	[XMIT_SZ_64BIT]		= 3,	\
+	[XMIT_SZ_256BIT]	= 5,	\
+}
+
+#define TS_INDEX2VAL(i)	(((i) & 7) << CHCR_TS_LOW_SHIFT)
+
+#endif /* CONFIG_CPU_SH4A */
+
+#endif
diff --git a/arch/sh/include/cpu-sh4/cpu/dma-sh4a.h b/arch/sh/include/cpu-sh4/cpu/dma-sh4a.h
index e734ea47d8a0..9647e681fd27 100644
--- a/arch/sh/include/cpu-sh4/cpu/dma-sh4a.h
+++ b/arch/sh/include/cpu-sh4/cpu/dma-sh4a.h
@@ -8,20 +8,12 @@
 #define DMAE0_IRQ	78	/* DMA Error IRQ*/
 #define SH_DMAC_BASE0	0xFE008020
 #define SH_DMARS_BASE0	0xFE009000
-#define CHCR_TS_LOW_MASK	0x00000018
-#define CHCR_TS_LOW_SHIFT	3
-#define CHCR_TS_HIGH_MASK	0
-#define CHCR_TS_HIGH_SHIFT	0
 #elif defined(CONFIG_CPU_SUBTYPE_SH7722)
 #define DMTE0_IRQ	48
 #define DMTE4_IRQ	76
 #define DMAE0_IRQ	78	/* DMA Error IRQ*/
 #define SH_DMAC_BASE0	0xFE008020
 #define SH_DMARS_BASE0	0xFE009000
-#define CHCR_TS_LOW_MASK	0x00000018
-#define CHCR_TS_LOW_SHIFT	3
-#define CHCR_TS_HIGH_MASK	0x00300000
-#define CHCR_TS_HIGH_SHIFT	20
 #elif defined(CONFIG_CPU_SUBTYPE_SH7763) || \
 	defined(CONFIG_CPU_SUBTYPE_SH7764)
 #define DMTE0_IRQ	34
@@ -29,10 +21,6 @@
 #define DMAE0_IRQ	38
 #define SH_DMAC_BASE0	0xFF608020
 #define SH_DMARS_BASE0	0xFF609000
-#define CHCR_TS_LOW_MASK	0x00000018
-#define CHCR_TS_LOW_SHIFT	3
-#define CHCR_TS_HIGH_MASK	0
-#define CHCR_TS_HIGH_SHIFT	0
 #elif defined(CONFIG_CPU_SUBTYPE_SH7723)
 #define DMTE0_IRQ	48	/* DMAC0A*/
 #define DMTE4_IRQ	76	/* DMAC0B */
@@ -46,10 +34,6 @@
 #define SH_DMAC_BASE0	0xFE008020
 #define SH_DMAC_BASE1	0xFDC08020
 #define SH_DMARS_BASE0	0xFDC09000
-#define CHCR_TS_LOW_MASK	0x00000018
-#define CHCR_TS_LOW_SHIFT	3
-#define CHCR_TS_HIGH_MASK	0
-#define CHCR_TS_HIGH_SHIFT	0
 #elif defined(CONFIG_CPU_SUBTYPE_SH7724)
 #define DMTE0_IRQ	48	/* DMAC0A*/
 #define DMTE4_IRQ	76	/* DMAC0B */
@@ -64,10 +48,6 @@
 #define SH_DMAC_BASE1	0xFDC08020
 #define SH_DMARS_BASE0	0xFE009000
 #define SH_DMARS_BASE1	0xFDC09000
-#define CHCR_TS_LOW_MASK	0x00000018
-#define CHCR_TS_LOW_SHIFT	3
-#define CHCR_TS_HIGH_MASK	0x00600000
-#define CHCR_TS_HIGH_SHIFT	21
 #elif defined(CONFIG_CPU_SUBTYPE_SH7780)
 #define DMTE0_IRQ	34
 #define DMTE4_IRQ	44
@@ -80,10 +60,6 @@
 #define SH_DMAC_BASE0	0xFC808020
 #define SH_DMAC_BASE1	0xFC818020
 #define SH_DMARS_BASE0	0xFC809000
-#define CHCR_TS_LOW_MASK	0x00000018
-#define CHCR_TS_LOW_SHIFT	3
-#define CHCR_TS_HIGH_MASK	0
-#define CHCR_TS_HIGH_SHIFT	0
 #else /* SH7785 */
 #define DMTE0_IRQ	33
 #define DMTE4_IRQ	37
@@ -97,10 +73,6 @@
 #define SH_DMAC_BASE0	0xFC808020
 #define SH_DMAC_BASE1	0xFCC08020
 #define SH_DMARS_BASE0	0xFC809000
-#define CHCR_TS_LOW_MASK	0x00000018
-#define CHCR_TS_LOW_SHIFT	3
-#define CHCR_TS_HIGH_MASK	0
-#define CHCR_TS_HIGH_SHIFT	0
 #endif
 
 #define REQ_HE		0x000000C0
@@ -108,38 +80,4 @@
 #define REQ_LE		0x00000040
 #define TM_BURST	0x00000020
 
-/*
- * The SuperH DMAC supports a number of transmit sizes, we list them here,
- * with their respective values as they appear in the CHCR registers.
- *
- * Defaults to a 64-bit transfer size.
- */
-enum {
-	XMIT_SZ_8BIT		= 0,
-	XMIT_SZ_16BIT		= 1,
-	XMIT_SZ_32BIT		= 2,
-	XMIT_SZ_64BIT		= 7,
-	XMIT_SZ_128BIT		= 3,
-	XMIT_SZ_256BIT		= 4,
-	XMIT_SZ_128BIT_BLK	= 0xb,
-	XMIT_SZ_256BIT_BLK	= 0xc,
-};
-
-/*
- * The DMA count is defined as the number of bytes to transfer.
- */
-#define TS_SHIFT {			\
-	[XMIT_SZ_8BIT]		= 0,	\
-	[XMIT_SZ_16BIT]		= 1,	\
-	[XMIT_SZ_32BIT]		= 2,	\
-	[XMIT_SZ_64BIT]		= 3,	\
-	[XMIT_SZ_128BIT]	= 4,	\
-	[XMIT_SZ_256BIT]	= 5,	\
-	[XMIT_SZ_128BIT_BLK]	= 4,	\
-	[XMIT_SZ_256BIT_BLK]	= 5,	\
-}
-
-#define TS_INDEX2VAL(i)	((((i) & 3) << CHCR_TS_LOW_SHIFT) | \
-			 ((((i) >> 2) & 3) << CHCR_TS_HIGH_SHIFT))
-
 #endif /* __ASM_SH_CPU_SH4_DMA_SH7780_H */
diff --git a/arch/sh/include/cpu-sh4/cpu/dma.h b/arch/sh/include/cpu-sh4/cpu/dma.h
index 114a369705bc..ca747e93c2ed 100644
--- a/arch/sh/include/cpu-sh4/cpu/dma.h
+++ b/arch/sh/include/cpu-sh4/cpu/dma.h
@@ -5,9 +5,8 @@
 
 #ifdef CONFIG_CPU_SH4A
 
-#define DMAOR_INIT	(DMAOR_DME)
-
 #include <cpu/dma-sh4a.h>
+
 #else /* CONFIG_CPU_SH4A */
 /*
  * SH7750/SH7751/SH7760
@@ -17,7 +16,6 @@
 #define DMTE6_IRQ	46
 #define DMAE0_IRQ	38
 
-#define DMAOR_INIT	(0x8000|DMAOR_DME)
 #define SH_DMAC_BASE0	0xffa00000
 #define SH_DMAC_BASE1	0xffa00070
 /* Definitions for the SuperH DMAC */
@@ -27,40 +25,8 @@
 #define TS_32		0x00000030
 #define TS_64		0x00000000
 
-#define CHCR_TS_LOW_MASK	0x70
-#define CHCR_TS_LOW_SHIFT	4
-#define CHCR_TS_HIGH_MASK	0
-#define CHCR_TS_HIGH_SHIFT	0
-
 #define DMAOR_COD	0x00000008
 
-/*
- * The SuperH DMAC supports a number of transmit sizes, we list them here,
- * with their respective values as they appear in the CHCR registers.
- *
- * Defaults to a 64-bit transfer size.
- */
-enum {
-	XMIT_SZ_8BIT	= 1,
-	XMIT_SZ_16BIT	= 2,
-	XMIT_SZ_32BIT	= 3,
-	XMIT_SZ_64BIT	= 0,
-	XMIT_SZ_256BIT	= 4,
-};
-
-/*
- * The DMA count is defined as the number of bytes to transfer.
- */
-#define TS_SHIFT {			\
-	[XMIT_SZ_8BIT]		= 0,	\
-	[XMIT_SZ_16BIT]		= 1,	\
-	[XMIT_SZ_32BIT]		= 2,	\
-	[XMIT_SZ_64BIT]		= 3,	\
-	[XMIT_SZ_256BIT]	= 5,	\
-}
-
-#define TS_INDEX2VAL(i)	(((i) & 7) << CHCR_TS_LOW_SHIFT)
-
 #endif
 
 #endif /* __ASM_CPU_SH4_DMA_H */
diff --git a/arch/sh/include/mach-migor/mach/migor.h b/arch/sh/include/mach-migor/mach/migor.h
index cee6cb88e020..42fccf93412e 100644
--- a/arch/sh/include/mach-migor/mach/migor.h
+++ b/arch/sh/include/mach-migor/mach/migor.h
@@ -1,6 +1,7 @@
 #ifndef __ASM_SH_MIGOR_H
 #define __ASM_SH_MIGOR_H
 
+#define PORT_MSELCRA 0xa4050180
 #define PORT_MSELCRB 0xa4050182
 #define BSC_CS4BCR 0xfec10010
 #define BSC_CS6ABCR 0xfec1001c
diff --git a/arch/sh/kernel/cpu/sh4a/setup-sh7722.c b/arch/sh/kernel/cpu/sh4a/setup-sh7722.c
index ef3f97827808..fd7e3639e845 100644
--- a/arch/sh/kernel/cpu/sh4a/setup-sh7722.c
+++ b/arch/sh/kernel/cpu/sh4a/setup-sh7722.c
@@ -7,19 +7,167 @@
  * License.  See the file "COPYING" in the main directory of this archive
  * for more details.
  */
-#include <linux/platform_device.h>
 #include <linux/init.h>
+#include <linux/mm.h>
+#include <linux/platform_device.h>
 #include <linux/serial.h>
 #include <linux/serial_sci.h>
-#include <linux/mm.h>
+#include <linux/sh_timer.h>
 #include <linux/uio_driver.h>
 #include <linux/usb/m66592.h>
-#include <linux/sh_timer.h>
+
 #include <asm/clock.h>
+#include <asm/dmaengine.h>
 #include <asm/mmzone.h>
-#include <asm/dma-sh.h>
+#include <asm/siu.h>
+
+#include <cpu/dma-register.h>
 #include <cpu/sh7722.h>
 
+static struct sh_dmae_slave_config sh7722_dmae_slaves[] = {
+	{
+		.slave_id	= SHDMA_SLAVE_SCIF0_TX,
+		.addr		= 0xffe0000c,
+		.chcr		= DM_FIX | SM_INC | 0x800 | TS_INDEX2VAL(XMIT_SZ_8BIT),
+		.mid_rid	= 0x21,
+	}, {
+		.slave_id	= SHDMA_SLAVE_SCIF0_RX,
+		.addr		= 0xffe00014,
+		.chcr		= DM_INC | SM_FIX | 0x800 | TS_INDEX2VAL(XMIT_SZ_8BIT),
+		.mid_rid	= 0x22,
+	}, {
+		.slave_id	= SHDMA_SLAVE_SCIF1_TX,
+		.addr		= 0xffe1000c,
+		.chcr		= DM_FIX | SM_INC | 0x800 | TS_INDEX2VAL(XMIT_SZ_8BIT),
+		.mid_rid	= 0x25,
+	}, {
+		.slave_id	= SHDMA_SLAVE_SCIF1_RX,
+		.addr		= 0xffe10014,
+		.chcr		= DM_INC | SM_FIX | 0x800 | TS_INDEX2VAL(XMIT_SZ_8BIT),
+		.mid_rid	= 0x26,
+	}, {
+		.slave_id	= SHDMA_SLAVE_SCIF2_TX,
+		.addr		= 0xffe2000c,
+		.chcr		= DM_FIX | SM_INC | 0x800 | TS_INDEX2VAL(XMIT_SZ_8BIT),
+		.mid_rid	= 0x29,
+	}, {
+		.slave_id	= SHDMA_SLAVE_SCIF2_RX,
+		.addr		= 0xffe20014,
+		.chcr		= DM_INC | SM_FIX | 0x800 | TS_INDEX2VAL(XMIT_SZ_8BIT),
+		.mid_rid	= 0x2a,
+	}, {
+		.slave_id	= SHDMA_SLAVE_SIUA_TX,
+		.addr		= 0xa454c098,
+		.chcr		= DM_FIX | SM_INC | 0x800 | TS_INDEX2VAL(XMIT_SZ_32BIT),
+		.mid_rid	= 0xb1,
+	}, {
+		.slave_id	= SHDMA_SLAVE_SIUA_RX,
+		.addr		= 0xa454c090,
+		.chcr		= DM_INC | SM_FIX | 0x800 | TS_INDEX2VAL(XMIT_SZ_32BIT),
+		.mid_rid	= 0xb2,
+	}, {
+		.slave_id	= SHDMA_SLAVE_SIUB_TX,
+		.addr		= 0xa454c09c,
+		.chcr		= DM_FIX | SM_INC | 0x800 | TS_INDEX2VAL(XMIT_SZ_32BIT),
+		.mid_rid	= 0xb5,
+	}, {
+		.slave_id	= SHDMA_SLAVE_SIUB_RX,
+		.addr		= 0xa454c094,
+		.chcr		= DM_INC | SM_FIX | 0x800 | TS_INDEX2VAL(XMIT_SZ_32BIT),
+		.mid_rid	= 0xb6,
+	},
+};
+
+static struct sh_dmae_channel sh7722_dmae_channels[] = {
+	{
+		.offset = 0,
+		.dmars = 0,
+		.dmars_bit = 0,
+	}, {
+		.offset = 0x10,
+		.dmars = 0,
+		.dmars_bit = 8,
+	}, {
+		.offset = 0x20,
+		.dmars = 4,
+		.dmars_bit = 0,
+	}, {
+		.offset = 0x30,
+		.dmars = 4,
+		.dmars_bit = 8,
+	}, {
+		.offset = 0x50,
+		.dmars = 8,
+		.dmars_bit = 0,
+	}, {
+		.offset = 0x60,
+		.dmars = 8,
+		.dmars_bit = 8,
+	}
+};
+
+static unsigned int ts_shift[] = TS_SHIFT;
+
+static struct sh_dmae_pdata dma_platform_data = {
+	.slave		= sh7722_dmae_slaves,
+	.slave_num	= ARRAY_SIZE(sh7722_dmae_slaves),
+	.channel	= sh7722_dmae_channels,
+	.channel_num	= ARRAY_SIZE(sh7722_dmae_channels),
+	.ts_low_shift	= CHCR_TS_LOW_SHIFT,
+	.ts_low_mask	= CHCR_TS_LOW_MASK,
+	.ts_high_shift	= CHCR_TS_HIGH_SHIFT,
+	.ts_high_mask	= CHCR_TS_HIGH_MASK,
+	.ts_shift	= ts_shift,
+	.ts_shift_num	= ARRAY_SIZE(ts_shift),
+	.dmaor_init	= DMAOR_INIT,
+};
+
+static struct resource sh7722_dmae_resources[] = {
+	[0] = {
+		/* Channel registers and DMAOR */
+		.start	= 0xfe008020,
+		.end	= 0xfe00808f,
+		.flags	= IORESOURCE_MEM,
+	},
+	[1] = {
+		/* DMARSx */
+		.start	= 0xfe009000,
+		.end	= 0xfe00900b,
+		.flags	= IORESOURCE_MEM,
+	},
+	{
+		/* DMA error IRQ */
+		.start	= 78,
+		.end	= 78,
+		.flags	= IORESOURCE_IRQ,
+	},
+	{
+		/* IRQ for channels 0-3 */
+		.start	= 48,
+		.end	= 51,
+		.flags	= IORESOURCE_IRQ,
+	},
+	{
+		/* IRQ for channels 4-5 */
+		.start	= 76,
+		.end	= 77,
+		.flags	= IORESOURCE_IRQ,
+	},
+};
+
+struct platform_device dma_device = {
+	.name		= "sh-dma-engine",
+	.id		= -1,
+	.resource	= sh7722_dmae_resources,
+	.num_resources	= ARRAY_SIZE(sh7722_dmae_resources),
+	.dev		= {
+		.platform_data	= &dma_platform_data,
+	},
+	.archdata = {
+		.hwblk_id = HWBLK_DMAC,
+	},
+};
+
 /* Serial */
 static struct plat_sci_port scif0_platform_data = {
 	.mapbase        = 0xffe00000,
@@ -388,15 +536,36 @@ static struct platform_device tmu2_device = {
 	},
 };
 
-static struct sh_dmae_pdata dma_platform_data = {
-	.mode = 0,
+static struct siu_platform siu_platform_data = {
+	.dma_dev	= &dma_device.dev,
+	.dma_slave_tx_a	= SHDMA_SLAVE_SIUA_TX,
+	.dma_slave_rx_a	= SHDMA_SLAVE_SIUA_RX,
+	.dma_slave_tx_b	= SHDMA_SLAVE_SIUB_TX,
+	.dma_slave_rx_b	= SHDMA_SLAVE_SIUB_RX,
 };
 
-static struct platform_device dma_device = {
-	.name		= "sh-dma-engine",
+static struct resource siu_resources[] = {
+	[0] = {
+		.start	= 0xa4540000,
+		.end	= 0xa454c10f,
+		.flags	= IORESOURCE_MEM,
+	},
+	[1] = {
+		.start	= 108,
+		.flags	= IORESOURCE_IRQ,
+	},
+};
+
+static struct platform_device siu_device = {
+	.name		= "sh_siu",
 	.id		= -1,
-	.dev		= {
-		.platform_data	= &dma_platform_data,
+	.dev = {
+		.platform_data	= &siu_platform_data,
+	},
+	.resource	= siu_resources,
+	.num_resources	= ARRAY_SIZE(siu_resources),
+	.archdata = {
+		.hwblk_id = HWBLK_SIU,
 	},
 };
 
@@ -414,6 +583,7 @@ static struct platform_device *sh7722_devices[] __initdata = {
 	&vpu_device,
 	&veu_device,
 	&jpu_device,
+	&siu_device,
 	&dma_device,
 };
 
diff --git a/arch/sh/kernel/cpu/sh4a/setup-sh7724.c b/arch/sh/kernel/cpu/sh4a/setup-sh7724.c
index 31e3451f7e3d..e7fa2a92fc1f 100644
--- a/arch/sh/kernel/cpu/sh4a/setup-sh7724.c
+++ b/arch/sh/kernel/cpu/sh4a/setup-sh7724.c
@@ -21,22 +21,189 @@
 #include <linux/sh_timer.h>
 #include <linux/io.h>
 #include <linux/notifier.h>
+
 #include <asm/suspend.h>
 #include <asm/clock.h>
-#include <asm/dma-sh.h>
+#include <asm/dmaengine.h>
 #include <asm/mmzone.h>
+
+#include <cpu/dma-register.h>
 #include <cpu/sh7724.h>
 
 /* DMA */
-static struct sh_dmae_pdata dma_platform_data = {
-	.mode = SHDMA_DMAOR1,
+static struct sh_dmae_channel sh7724_dmae0_channels[] = {
+	{
+		.offset = 0,
+		.dmars = 0,
+		.dmars_bit = 0,
+	}, {
+		.offset = 0x10,
+		.dmars = 0,
+		.dmars_bit = 8,
+	}, {
+		.offset = 0x20,
+		.dmars = 4,
+		.dmars_bit = 0,
+	}, {
+		.offset = 0x30,
+		.dmars = 4,
+		.dmars_bit = 8,
+	}, {
+		.offset = 0x50,
+		.dmars = 8,
+		.dmars_bit = 0,
+	}, {
+		.offset = 0x60,
+		.dmars = 8,
+		.dmars_bit = 8,
+	}
+};
+
+static struct sh_dmae_channel sh7724_dmae1_channels[] = {
+	{
+		.offset = 0,
+		.dmars = 0,
+		.dmars_bit = 0,
+	}, {
+		.offset = 0x10,
+		.dmars = 0,
+		.dmars_bit = 8,
+	}, {
+		.offset = 0x20,
+		.dmars = 4,
+		.dmars_bit = 0,
+	}, {
+		.offset = 0x30,
+		.dmars = 4,
+		.dmars_bit = 8,
+	}, {
+		.offset = 0x50,
+		.dmars = 8,
+		.dmars_bit = 0,
+	}, {
+		.offset = 0x60,
+		.dmars = 8,
+		.dmars_bit = 8,
+	}
+};
+
+static unsigned int ts_shift[] = TS_SHIFT;
+
+static struct sh_dmae_pdata dma0_platform_data = {
+	.channel	= sh7724_dmae0_channels,
+	.channel_num	= ARRAY_SIZE(sh7724_dmae0_channels),
+	.ts_low_shift	= CHCR_TS_LOW_SHIFT,
+	.ts_low_mask	= CHCR_TS_LOW_MASK,
+	.ts_high_shift	= CHCR_TS_HIGH_SHIFT,
+	.ts_high_mask	= CHCR_TS_HIGH_MASK,
+	.ts_shift	= ts_shift,
+	.ts_shift_num	= ARRAY_SIZE(ts_shift),
+	.dmaor_init	= DMAOR_INIT,
+};
+
+static struct sh_dmae_pdata dma1_platform_data = {
+	.channel	= sh7724_dmae1_channels,
+	.channel_num	= ARRAY_SIZE(sh7724_dmae1_channels),
+	.ts_low_shift	= CHCR_TS_LOW_SHIFT,
+	.ts_low_mask	= CHCR_TS_LOW_MASK,
+	.ts_high_shift	= CHCR_TS_HIGH_SHIFT,
+	.ts_high_mask	= CHCR_TS_HIGH_MASK,
+	.ts_shift	= ts_shift,
+	.ts_shift_num	= ARRAY_SIZE(ts_shift),
+	.dmaor_init	= DMAOR_INIT,
+};
+
+/* Resource order important! */
+static struct resource sh7724_dmae0_resources[] = {
+	{
+		/* Channel registers and DMAOR */
+		.start	= 0xfe008020,
+		.end	= 0xfe00808f,
+		.flags	= IORESOURCE_MEM,
+	},
+	{
+		/* DMARSx */
+		.start	= 0xfe009000,
+		.end	= 0xfe00900b,
+		.flags	= IORESOURCE_MEM,
+	},
+	{
+		/* DMA error IRQ */
+		.start	= 78,
+		.end	= 78,
+		.flags	= IORESOURCE_IRQ,
+	},
+	{
+		/* IRQ for channels 0-3 */
+		.start	= 48,
+		.end	= 51,
+		.flags	= IORESOURCE_IRQ,
+	},
+	{
+		/* IRQ for channels 4-5 */
+		.start	= 76,
+		.end	= 77,
+		.flags	= IORESOURCE_IRQ,
+	},
 };
 
-static struct platform_device dma_device = {
-	.name	= "sh-dma-engine",
-	.id		= -1,
-	.dev	= {
-		.platform_data	= &dma_platform_data,
+/* Resource order important! */
+static struct resource sh7724_dmae1_resources[] = {
+	{
+		/* Channel registers and DMAOR */
+		.start	= 0xfdc08020,
+		.end	= 0xfdc0808f,
+		.flags	= IORESOURCE_MEM,
+	},
+	{
+		/* DMARSx */
+		.start	= 0xfdc09000,
+		.end	= 0xfdc0900b,
+		.flags	= IORESOURCE_MEM,
+	},
+	{
+		/* DMA error IRQ */
+		.start	= 74,
+		.end	= 74,
+		.flags	= IORESOURCE_IRQ,
+	},
+	{
+		/* IRQ for channels 0-3 */
+		.start	= 40,
+		.end	= 43,
+		.flags	= IORESOURCE_IRQ,
+	},
+	{
+		/* IRQ for channels 4-5 */
+		.start	= 72,
+		.end	= 73,
+		.flags	= IORESOURCE_IRQ,
+	},
+};
+
+static struct platform_device dma0_device = {
+	.name		= "sh-dma-engine",
+	.id		= 0,
+	.resource	= sh7724_dmae0_resources,
+	.num_resources	= ARRAY_SIZE(sh7724_dmae0_resources),
+	.dev		= {
+		.platform_data	= &dma0_platform_data,
+	},
+	.archdata = {
+		.hwblk_id = HWBLK_DMAC0,
+	},
+};
+
+static struct platform_device dma1_device = {
+	.name		= "sh-dma-engine",
+	.id		= 1,
+	.resource	= sh7724_dmae1_resources,
+	.num_resources	= ARRAY_SIZE(sh7724_dmae1_resources),
+	.dev		= {
+		.platform_data	= &dma1_platform_data,
+	},
+	.archdata = {
+		.hwblk_id = HWBLK_DMAC1,
 	},
 };
 
@@ -663,7 +830,8 @@ static struct platform_device *sh7724_devices[] __initdata = {
 	&tmu3_device,
 	&tmu4_device,
 	&tmu5_device,
-	&dma_device,
+	&dma0_device,
+	&dma1_device,
 	&rtc_device,
 	&iic0_device,
 	&iic1_device,
diff --git a/arch/sh/kernel/cpu/sh4a/setup-sh7780.c b/arch/sh/kernel/cpu/sh4a/setup-sh7780.c
index f8f21618d785..02e792c90de6 100644
--- a/arch/sh/kernel/cpu/sh4a/setup-sh7780.c
+++ b/arch/sh/kernel/cpu/sh4a/setup-sh7780.c
@@ -13,7 +13,10 @@
 #include <linux/io.h>
 #include <linux/serial_sci.h>
 #include <linux/sh_timer.h>
-#include <asm/dma-sh.h>
+
+#include <asm/dmaengine.h>
+
+#include <cpu/dma-register.h>
 
 static struct plat_sci_port scif0_platform_data = {
 	.mapbase	= 0xffe00000,
@@ -247,15 +250,131 @@ static struct platform_device rtc_device = {
 	.resource	= rtc_resources,
 };
 
-static struct sh_dmae_pdata dma_platform_data = {
-	.mode = (SHDMA_MIX_IRQ | SHDMA_DMAOR1),
+/* DMA */
+static struct sh_dmae_channel sh7780_dmae0_channels[] = {
+	{
+		.offset = 0,
+		.dmars = 0,
+		.dmars_bit = 0,
+	}, {
+		.offset = 0x10,
+		.dmars = 0,
+		.dmars_bit = 8,
+	}, {
+		.offset = 0x20,
+		.dmars = 4,
+		.dmars_bit = 0,
+	}, {
+		.offset = 0x30,
+		.dmars = 4,
+		.dmars_bit = 8,
+	}, {
+		.offset = 0x50,
+		.dmars = 8,
+		.dmars_bit = 0,
+	}, {
+		.offset = 0x60,
+		.dmars = 8,
+		.dmars_bit = 8,
+	}
+};
+
+static struct sh_dmae_channel sh7780_dmae1_channels[] = {
+	{
+		.offset = 0,
+	}, {
+		.offset = 0x10,
+	}, {
+		.offset = 0x20,
+	}, {
+		.offset = 0x30,
+	}, {
+		.offset = 0x50,
+	}, {
+		.offset = 0x60,
+	}
+};
+
+static unsigned int ts_shift[] = TS_SHIFT;
+
+static struct sh_dmae_pdata dma0_platform_data = {
+	.channel	= sh7780_dmae0_channels,
+	.channel_num	= ARRAY_SIZE(sh7780_dmae0_channels),
+	.ts_low_shift	= CHCR_TS_LOW_SHIFT,
+	.ts_low_mask	= CHCR_TS_LOW_MASK,
+	.ts_high_shift	= CHCR_TS_HIGH_SHIFT,
+	.ts_high_mask	= CHCR_TS_HIGH_MASK,
+	.ts_shift	= ts_shift,
+	.ts_shift_num	= ARRAY_SIZE(ts_shift),
+	.dmaor_init	= DMAOR_INIT,
+};
+
+static struct sh_dmae_pdata dma1_platform_data = {
+	.channel	= sh7780_dmae1_channels,
+	.channel_num	= ARRAY_SIZE(sh7780_dmae1_channels),
+	.ts_low_shift	= CHCR_TS_LOW_SHIFT,
+	.ts_low_mask	= CHCR_TS_LOW_MASK,
+	.ts_high_shift	= CHCR_TS_HIGH_SHIFT,
+	.ts_high_mask	= CHCR_TS_HIGH_MASK,
+	.ts_shift	= ts_shift,
+	.ts_shift_num	= ARRAY_SIZE(ts_shift),
+	.dmaor_init	= DMAOR_INIT,
 };
 
-static struct platform_device dma_device = {
+static struct resource sh7780_dmae0_resources[] = {
+	[0] = {
+		/* Channel registers and DMAOR */
+		.start	= 0xfc808020,
+		.end	= 0xfc80808f,
+		.flags	= IORESOURCE_MEM,
+	},
+	[1] = {
+		/* DMARSx */
+		.start	= 0xfc809000,
+		.end	= 0xfc80900b,
+		.flags	= IORESOURCE_MEM,
+	},
+	{
+		/* Real DMA error IRQ is 38, and channel IRQs are 34-37, 44-45 */
+		.start	= 34,
+		.end	= 34,
+		.flags	= IORESOURCE_IRQ | IORESOURCE_IRQ_SHAREABLE,
+	},
+};
+
+static struct resource sh7780_dmae1_resources[] = {
+	[0] = {
+		/* Channel registers and DMAOR */
+		.start	= 0xfc818020,
+		.end	= 0xfc81808f,
+		.flags	= IORESOURCE_MEM,
+	},
+	/* DMAC1 has no DMARS */
+	{
+		/* Real DMA error IRQ is 38, and channel IRQs are 46-47, 92-95 */
+		.start	= 46,
+		.end	= 46,
+		.flags	= IORESOURCE_IRQ | IORESOURCE_IRQ_SHAREABLE,
+	},
+};
+
+static struct platform_device dma0_device = {
 	.name           = "sh-dma-engine",
-	.id             = -1,
+	.id             = 0,
+	.resource	= sh7780_dmae0_resources,
+	.num_resources	= ARRAY_SIZE(sh7780_dmae0_resources),
 	.dev            = {
-		.platform_data  = &dma_platform_data,
+		.platform_data	= &dma0_platform_data,
+	},
+};
+
+static struct platform_device dma1_device = {
+	.name		= "sh-dma-engine",
+	.id		= 1,
+	.resource	= sh7780_dmae1_resources,
+	.num_resources	= ARRAY_SIZE(sh7780_dmae1_resources),
+	.dev		= {
+		.platform_data	= &dma1_platform_data,
 	},
 };
 
@@ -269,7 +388,8 @@ static struct platform_device *sh7780_devices[] __initdata = {
 	&tmu4_device,
 	&tmu5_device,
 	&rtc_device,
-	&dma_device,
+	&dma0_device,
+	&dma1_device,
 };
 
 static int __init sh7780_devices_setup(void)
diff --git a/arch/sh/kernel/cpu/sh4a/setup-sh7785.c b/arch/sh/kernel/cpu/sh4a/setup-sh7785.c
index 23448d8c6711..1fcd88b1671e 100644
--- a/arch/sh/kernel/cpu/sh4a/setup-sh7785.c
+++ b/arch/sh/kernel/cpu/sh4a/setup-sh7785.c
@@ -14,9 +14,12 @@
 #include <linux/io.h>
 #include <linux/mm.h>
 #include <linux/sh_timer.h>
-#include <asm/dma-sh.h>
+
+#include <asm/dmaengine.h>
 #include <asm/mmzone.h>
 
+#include <cpu/dma-register.h>
+
 static struct plat_sci_port scif0_platform_data = {
 	.mapbase	= 0xffea0000,
 	.flags		= UPF_BOOT_AUTOCONF,
@@ -295,15 +298,131 @@ static struct platform_device tmu5_device = {
 	.num_resources	= ARRAY_SIZE(tmu5_resources),
 };
 
-static struct sh_dmae_pdata dma_platform_data = {
-	.mode = (SHDMA_MIX_IRQ | SHDMA_DMAOR1),
+/* DMA */
+static struct sh_dmae_channel sh7785_dmae0_channels[] = {
+	{
+		.offset = 0,
+		.dmars = 0,
+		.dmars_bit = 0,
+	}, {
+		.offset = 0x10,
+		.dmars = 0,
+		.dmars_bit = 8,
+	}, {
+		.offset = 0x20,
+		.dmars = 4,
+		.dmars_bit = 0,
+	}, {
+		.offset = 0x30,
+		.dmars = 4,
+		.dmars_bit = 8,
+	}, {
+		.offset = 0x50,
+		.dmars = 8,
+		.dmars_bit = 0,
+	}, {
+		.offset = 0x60,
+		.dmars = 8,
+		.dmars_bit = 8,
+	}
+};
+
+static struct sh_dmae_channel sh7785_dmae1_channels[] = {
+	{
+		.offset = 0,
+	}, {
+		.offset = 0x10,
+	}, {
+		.offset = 0x20,
+	}, {
+		.offset = 0x30,
+	}, {
+		.offset = 0x50,
+	}, {
+		.offset = 0x60,
+	}
+};
+
+static unsigned int ts_shift[] = TS_SHIFT;
+
+static struct sh_dmae_pdata dma0_platform_data = {
+	.channel	= sh7785_dmae0_channels,
+	.channel_num	= ARRAY_SIZE(sh7785_dmae0_channels),
+	.ts_low_shift	= CHCR_TS_LOW_SHIFT,
+	.ts_low_mask	= CHCR_TS_LOW_MASK,
+	.ts_high_shift	= CHCR_TS_HIGH_SHIFT,
+	.ts_high_mask	= CHCR_TS_HIGH_MASK,
+	.ts_shift	= ts_shift,
+	.ts_shift_num	= ARRAY_SIZE(ts_shift),
+	.dmaor_init	= DMAOR_INIT,
+};
+
+static struct sh_dmae_pdata dma1_platform_data = {
+	.channel	= sh7785_dmae1_channels,
+	.channel_num	= ARRAY_SIZE(sh7785_dmae1_channels),
+	.ts_low_shift	= CHCR_TS_LOW_SHIFT,
+	.ts_low_mask	= CHCR_TS_LOW_MASK,
+	.ts_high_shift	= CHCR_TS_HIGH_SHIFT,
+	.ts_high_mask	= CHCR_TS_HIGH_MASK,
+	.ts_shift	= ts_shift,
+	.ts_shift_num	= ARRAY_SIZE(ts_shift),
+	.dmaor_init	= DMAOR_INIT,
 };
 
-static struct platform_device dma_device = {
+static struct resource sh7785_dmae0_resources[] = {
+	[0] = {
+		/* Channel registers and DMAOR */
+		.start	= 0xfc808020,
+		.end	= 0xfc80808f,
+		.flags	= IORESOURCE_MEM,
+	},
+	[1] = {
+		/* DMARSx */
+		.start	= 0xfc809000,
+		.end	= 0xfc80900b,
+		.flags	= IORESOURCE_MEM,
+	},
+	{
+		/* Real DMA error IRQ is 39, and channel IRQs are 33-38 */
+		.start	= 33,
+		.end	= 33,
+		.flags	= IORESOURCE_IRQ | IORESOURCE_IRQ_SHAREABLE,
+	},
+};
+
+static struct resource sh7785_dmae1_resources[] = {
+	[0] = {
+		/* Channel registers and DMAOR */
+		.start	= 0xfcc08020,
+		.end	= 0xfcc0808f,
+		.flags	= IORESOURCE_MEM,
+	},
+	/* DMAC1 has no DMARS */
+	{
+		/* Real DMA error IRQ is 58, and channel IRQs are 52-57 */
+		.start	= 52,
+		.end	= 52,
+		.flags	= IORESOURCE_IRQ | IORESOURCE_IRQ_SHAREABLE,
+	},
+};
+
+static struct platform_device dma0_device = {
 	.name           = "sh-dma-engine",
-	.id             = -1,
+	.id             = 0,
+	.resource	= sh7785_dmae0_resources,
+	.num_resources	= ARRAY_SIZE(sh7785_dmae0_resources),
 	.dev            = {
-		.platform_data  = &dma_platform_data,
+		.platform_data	= &dma0_platform_data,
+	},
+};
+
+static struct platform_device dma1_device = {
+	.name		= "sh-dma-engine",
+	.id		= 1,
+	.resource	= sh7785_dmae1_resources,
+	.num_resources	= ARRAY_SIZE(sh7785_dmae1_resources),
+	.dev		= {
+		.platform_data	= &dma1_platform_data,
 	},
 };
 
@@ -320,7 +439,8 @@ static struct platform_device *sh7785_devices[] __initdata = {
 	&tmu3_device,
 	&tmu4_device,
 	&tmu5_device,
-	&dma_device,
+	&dma0_device,
+	&dma1_device,
 };
 
 static int __init sh7785_devices_setup(void)
diff --git a/arch/sh/kernel/hw_breakpoint.c b/arch/sh/kernel/hw_breakpoint.c
index e2f1753d275c..675eea7785d9 100644
--- a/arch/sh/kernel/hw_breakpoint.c
+++ b/arch/sh/kernel/hw_breakpoint.c
@@ -143,26 +143,6 @@ static int arch_check_va_in_kernelspace(unsigned long va, u8 hbp_len)
 	return (va >= TASK_SIZE) && ((va + len - 1) >= TASK_SIZE);
 }
 
-/*
- * Store a breakpoint's encoded address, length, and type.
- */
-static int arch_store_info(struct perf_event *bp)
-{
-	struct arch_hw_breakpoint *info = counter_arch_bp(bp);
-
-	/*
-	 * User-space requests will always have the address field populated
-	 * For kernel-addresses, either the address or symbol name can be
-	 * specified.
-	 */
-	if (info->name)
-		info->address = (unsigned long)kallsyms_lookup_name(info->name);
-	if (info->address)
-		return 0;
-
-	return -EINVAL;
-}
-
 int arch_bp_generic_fields(int sh_len, int sh_type,
 			   int *gen_len, int *gen_type)
 {
@@ -276,10 +256,12 @@ int arch_validate_hwbkpt_settings(struct perf_event *bp,
 		return ret;
 	}
 
-	ret = arch_store_info(bp);
-
-	if (ret < 0)
-		return ret;
+	/*
+	 * For kernel-addresses, either the address or symbol name can be
+	 * specified.
+	 */
+	if (info->name)
+		info->address = (unsigned long)kallsyms_lookup_name(info->name);
 
 	/*
 	 * Check that the low-order bits of the address are appropriate
diff --git a/arch/sh/kernel/setup.c b/arch/sh/kernel/setup.c
index 3459e70eed72..8870d6ba64bf 100644
--- a/arch/sh/kernel/setup.c
+++ b/arch/sh/kernel/setup.c
@@ -443,7 +443,7 @@ void __init setup_arch(char **cmdline_p)
 
 	nodes_clear(node_online_map);
 
-	/* Setup bootmem with available RAM */
+	pmb_init();
 	lmb_init();
 	setup_memory();
 	sparse_init();
@@ -452,7 +452,6 @@ void __init setup_arch(char **cmdline_p)
 	conswitchp = &dummy_con;
 #endif
 	paging_init();
-	pmb_init();
 
 	ioremap_fixed_init();
 
diff --git a/arch/sh/kernel/time.c b/arch/sh/kernel/time.c
index 953fa1613312..8a0072de2bcc 100644
--- a/arch/sh/kernel/time.c
+++ b/arch/sh/kernel/time.c
@@ -39,12 +39,12 @@ static int null_rtc_set_time(const time_t secs)
 void (*rtc_sh_get_time)(struct timespec *) = null_rtc_get_time;
 int (*rtc_sh_set_time)(const time_t) = null_rtc_set_time;
 
-#ifdef CONFIG_GENERIC_CMOS_UPDATE
 void read_persistent_clock(struct timespec *ts)
 {
 	rtc_sh_get_time(ts);
 }
 
+#ifdef CONFIG_GENERIC_CMOS_UPDATE
 int update_persistent_clock(struct timespec now)
 {
 	return rtc_sh_set_time(now.tv_sec);
@@ -113,9 +113,5 @@ void __init time_init(void)
 	hwblk_init();
 	clk_init();
 
-	rtc_sh_get_time(&xtime);
-	set_normalized_timespec(&wall_to_monotonic,
-				-xtime.tv_sec, -xtime.tv_nsec);
-
 	late_time_init = sh_late_time_init;
 }
diff --git a/arch/sh/lib/libgcc.h b/arch/sh/lib/libgcc.h
index 3f19d1c5d942..05909d58e2fe 100644
--- a/arch/sh/lib/libgcc.h
+++ b/arch/sh/lib/libgcc.h
@@ -17,8 +17,7 @@ struct DWstruct {
 #error I feel sick.
 #endif
 
-typedef union
-{
+typedef union {
 	struct DWstruct s;
 	long long ll;
 } DWunion;
diff --git a/arch/sh/mm/ioremap.c b/arch/sh/mm/ioremap.c
index c68d2d7d00a9..1ab2385ecefe 100644
--- a/arch/sh/mm/ioremap.c
+++ b/arch/sh/mm/ioremap.c
@@ -34,11 +34,12 @@
  * caller shouldn't need to know that small detail.
  */
 void __iomem * __init_refok
-__ioremap_caller(unsigned long phys_addr, unsigned long size,
+__ioremap_caller(phys_addr_t phys_addr, unsigned long size,
 		 pgprot_t pgprot, void *caller)
 {
 	struct vm_struct *area;
 	unsigned long offset, last_addr, addr, orig_addr;
+	void __iomem *mapped;
 
 	/* Don't allow wraparound or zero size */
 	last_addr = phys_addr + size - 1;
@@ -46,6 +47,20 @@ __ioremap_caller(unsigned long phys_addr, unsigned long size,
 		return NULL;
 
 	/*
+	 * If we can't yet use the regular approach, go the fixmap route.
+	 */
+	if (!mem_init_done)
+		return ioremap_fixed(phys_addr, size, pgprot);
+
+	/*
+	 * First try to remap through the PMB.
+	 * PMB entries are all pre-faulted.
+	 */
+	mapped = pmb_remap_caller(phys_addr, size, pgprot, caller);
+	if (mapped && !IS_ERR(mapped))
+		return mapped;
+
+	/*
 	 * Mappings have to be page-aligned
 	 */
 	offset = phys_addr & ~PAGE_MASK;
@@ -53,12 +68,6 @@ __ioremap_caller(unsigned long phys_addr, unsigned long size,
 	size = PAGE_ALIGN(last_addr+1) - phys_addr;
 
 	/*
-	 * If we can't yet use the regular approach, go the fixmap route.
-	 */
-	if (!mem_init_done)
-		return ioremap_fixed(phys_addr, offset, size, pgprot);
-
-	/*
 	 * Ok, go for it..
 	 */
 	area = get_vm_area_caller(size, VM_IOREMAP, caller);
@@ -67,33 +76,10 @@ __ioremap_caller(unsigned long phys_addr, unsigned long size,
 	area->phys_addr = phys_addr;
 	orig_addr = addr = (unsigned long)area->addr;
 
-#ifdef CONFIG_PMB
-	/*
-	 * First try to remap through the PMB once a valid VMA has been
-	 * established. Smaller allocations (or the rest of the size
-	 * remaining after a PMB mapping due to the size not being
-	 * perfectly aligned on a PMB size boundary) are then mapped
-	 * through the UTLB using conventional page tables.
-	 *
-	 * PMB entries are all pre-faulted.
-	 */
-	if (unlikely(phys_addr >= P1SEG)) {
-		unsigned long mapped;
-
-		mapped = pmb_remap(addr, phys_addr, size, pgprot);
-		if (likely(mapped)) {
-			addr		+= mapped;
-			phys_addr	+= mapped;
-			size		-= mapped;
-		}
+	if (ioremap_page_range(addr, addr + size, phys_addr, pgprot)) {
+		vunmap((void *)orig_addr);
+		return NULL;
 	}
-#endif
-
-	if (likely(size))
-		if (ioremap_page_range(addr, addr + size, phys_addr, pgprot)) {
-			vunmap((void *)orig_addr);
-			return NULL;
-		}
 
 	return (void __iomem *)(offset + (char *)orig_addr);
 }
@@ -133,23 +119,11 @@ void __iounmap(void __iomem *addr)
 	if (iounmap_fixed(addr) == 0)
 		return;
 
-#ifdef CONFIG_PMB
 	/*
-	 * Purge any PMB entries that may have been established for this
-	 * mapping, then proceed with conventional VMA teardown.
-	 *
-	 * XXX: Note that due to the way that remove_vm_area() does
-	 * matching of the resultant VMA, we aren't able to fast-forward
-	 * the address past the PMB space until the end of the VMA where
-	 * the page tables reside. As such, unmap_vm_area() will be
-	 * forced to linearly scan over the area until it finds the page
-	 * tables where PTEs that need to be unmapped actually reside,
-	 * which is far from optimal. Perhaps we need to use a separate
-	 * VMA for the PMB mappings?
-	 *					-- PFM.
+	 * If the PMB handled it, there's nothing else to do.
 	 */
-	pmb_unmap(vaddr);
-#endif
+	if (pmb_unmap(addr) == 0)
+		return;
 
 	p = remove_vm_area((void *)(vaddr & PAGE_MASK));
 	if (!p) {
diff --git a/arch/sh/mm/ioremap_fixed.c b/arch/sh/mm/ioremap_fixed.c
index 0b78b1e20ef1..7f682e5dafcf 100644
--- a/arch/sh/mm/ioremap_fixed.c
+++ b/arch/sh/mm/ioremap_fixed.c
@@ -45,14 +45,21 @@ void __init ioremap_fixed_init(void)
 }
 
 void __init __iomem *
-ioremap_fixed(resource_size_t phys_addr, unsigned long offset,
-	      unsigned long size, pgprot_t prot)
+ioremap_fixed(phys_addr_t phys_addr, unsigned long size, pgprot_t prot)
 {
 	enum fixed_addresses idx0, idx;
 	struct ioremap_map *map;
 	unsigned int nrpages;
+	unsigned long offset;
 	int i, slot;
 
+	/*
+	 * Mappings have to be page-aligned
+	 */
+	offset = phys_addr & ~PAGE_MASK;
+	phys_addr &= PAGE_MASK;
+	size = PAGE_ALIGN(phys_addr + size) - phys_addr;
+
 	slot = -1;
 	for (i = 0; i < FIX_N_IOREMAPS; i++) {
 		map = &ioremap_maps[i];
diff --git a/arch/sh/mm/numa.c b/arch/sh/mm/numa.c
index 422e92721878..961b34085e3b 100644
--- a/arch/sh/mm/numa.c
+++ b/arch/sh/mm/numa.c
@@ -74,6 +74,9 @@ void __init setup_bootmem_node(int nid, unsigned long start, unsigned long end)
 	start_pfn = start >> PAGE_SHIFT;
 	end_pfn = end >> PAGE_SHIFT;
 
+	pmb_bolt_mapping((unsigned long)__va(start), start, end - start,
+			 PAGE_KERNEL);
+
 	lmb_add(start, end - start);
 
 	__add_active_range(nid, start_pfn, end_pfn);
diff --git a/arch/sh/mm/pmb.c b/arch/sh/mm/pmb.c
index 198bcff5e96f..a4662e2782c3 100644
--- a/arch/sh/mm/pmb.c
+++ b/arch/sh/mm/pmb.c
@@ -23,7 +23,8 @@
 #include <linux/err.h>
 #include <linux/io.h>
 #include <linux/spinlock.h>
-#include <linux/rwlock.h>
+#include <linux/vmalloc.h>
+#include <asm/cacheflush.h>
 #include <asm/sizes.h>
 #include <asm/system.h>
 #include <asm/uaccess.h>
@@ -52,12 +53,24 @@ struct pmb_entry {
 	struct pmb_entry *link;
 };
 
+static struct {
+	unsigned long size;
+	int flag;
+} pmb_sizes[] = {
+	{ .size	= SZ_512M, .flag = PMB_SZ_512M, },
+	{ .size = SZ_128M, .flag = PMB_SZ_128M, },
+	{ .size = SZ_64M,  .flag = PMB_SZ_64M,  },
+	{ .size = SZ_16M,  .flag = PMB_SZ_16M,  },
+};
+
 static void pmb_unmap_entry(struct pmb_entry *, int depth);
 
 static DEFINE_RWLOCK(pmb_rwlock);
 static struct pmb_entry pmb_entry_list[NR_PMB_ENTRIES];
 static DECLARE_BITMAP(pmb_map, NR_PMB_ENTRIES);
 
+static unsigned int pmb_iomapping_enabled;
+
 static __always_inline unsigned long mk_pmb_entry(unsigned int entry)
 {
 	return (entry & PMB_E_MASK) << PMB_E_SHIFT;
@@ -73,6 +86,142 @@ static __always_inline unsigned long mk_pmb_data(unsigned int entry)
 	return mk_pmb_entry(entry) | PMB_DATA;
 }
 
+static __always_inline unsigned int pmb_ppn_in_range(unsigned long ppn)
+{
+	return ppn >= __pa(memory_start) && ppn < __pa(memory_end);
+}
+
+/*
+ * Ensure that the PMB entries match our cache configuration.
+ *
+ * When we are in 32-bit address extended mode, CCR.CB becomes
+ * invalid, so care must be taken to manually adjust cacheable
+ * translations.
+ */
+static __always_inline unsigned long pmb_cache_flags(void)
+{
+	unsigned long flags = 0;
+
+#if defined(CONFIG_CACHE_OFF)
+	flags |= PMB_WT | PMB_UB;
+#elif defined(CONFIG_CACHE_WRITETHROUGH)
+	flags |= PMB_C | PMB_WT | PMB_UB;
+#elif defined(CONFIG_CACHE_WRITEBACK)
+	flags |= PMB_C;
+#endif
+
+	return flags;
+}
+
+/*
+ * Convert typical pgprot value to the PMB equivalent
+ */
+static inline unsigned long pgprot_to_pmb_flags(pgprot_t prot)
+{
+	unsigned long pmb_flags = 0;
+	u64 flags = pgprot_val(prot);
+
+	if (flags & _PAGE_CACHABLE)
+		pmb_flags |= PMB_C;
+	if (flags & _PAGE_WT)
+		pmb_flags |= PMB_WT | PMB_UB;
+
+	return pmb_flags;
+}
+
+static inline bool pmb_can_merge(struct pmb_entry *a, struct pmb_entry *b)
+{
+	return (b->vpn == (a->vpn + a->size)) &&
+	       (b->ppn == (a->ppn + a->size)) &&
+	       (b->flags == a->flags);
+}
+
+static bool pmb_mapping_exists(unsigned long vaddr, phys_addr_t phys,
+			       unsigned long size)
+{
+	int i;
+
+	read_lock(&pmb_rwlock);
+
+	for (i = 0; i < ARRAY_SIZE(pmb_entry_list); i++) {
+		struct pmb_entry *pmbe, *iter;
+		unsigned long span;
+
+		if (!test_bit(i, pmb_map))
+			continue;
+
+		pmbe = &pmb_entry_list[i];
+
+		/*
+		 * See if VPN and PPN are bounded by an existing mapping.
+		 */
+		if ((vaddr < pmbe->vpn) || (vaddr >= (pmbe->vpn + pmbe->size)))
+			continue;
+		if ((phys < pmbe->ppn) || (phys >= (pmbe->ppn + pmbe->size)))
+			continue;
+
+		/*
+		 * Now see if we're in range of a simple mapping.
+		 */
+		if (size <= pmbe->size) {
+			read_unlock(&pmb_rwlock);
+			return true;
+		}
+
+		span = pmbe->size;
+
+		/*
+		 * Finally for sizes that involve compound mappings, walk
+		 * the chain.
+		 */
+		for (iter = pmbe->link; iter; iter = iter->link)
+			span += iter->size;
+
+		/*
+		 * Nothing else to do if the range requirements are met.
+		 */
+		if (size <= span) {
+			read_unlock(&pmb_rwlock);
+			return true;
+		}
+	}
+
+	read_unlock(&pmb_rwlock);
+	return false;
+}
+
+static bool pmb_size_valid(unsigned long size)
+{
+	int i;
+
+	for (i = 0; i < ARRAY_SIZE(pmb_sizes); i++)
+		if (pmb_sizes[i].size == size)
+			return true;
+
+	return false;
+}
+
+static inline bool pmb_addr_valid(unsigned long addr, unsigned long size)
+{
+	return (addr >= P1SEG && (addr + size - 1) < P3SEG);
+}
+
+static inline bool pmb_prot_valid(pgprot_t prot)
+{
+	return (pgprot_val(prot) & _PAGE_USER) == 0;
+}
+
+static int pmb_size_to_flags(unsigned long size)
+{
+	int i;
+
+	for (i = 0; i < ARRAY_SIZE(pmb_sizes); i++)
+		if (pmb_sizes[i].size == size)
+			return pmb_sizes[i].flag;
+
+	return 0;
+}
+
 static int pmb_alloc_entry(void)
 {
 	int pos;
@@ -140,33 +289,22 @@ static void pmb_free(struct pmb_entry *pmbe)
 }
 
 /*
- * Ensure that the PMB entries match our cache configuration.
- *
- * When we are in 32-bit address extended mode, CCR.CB becomes
- * invalid, so care must be taken to manually adjust cacheable
- * translations.
+ * Must be run uncached.
  */
-static __always_inline unsigned long pmb_cache_flags(void)
+static void __set_pmb_entry(struct pmb_entry *pmbe)
 {
-	unsigned long flags = 0;
+	unsigned long addr, data;
 
-#if defined(CONFIG_CACHE_WRITETHROUGH)
-	flags |= PMB_C | PMB_WT | PMB_UB;
-#elif defined(CONFIG_CACHE_WRITEBACK)
-	flags |= PMB_C;
-#endif
+	addr = mk_pmb_addr(pmbe->entry);
+	data = mk_pmb_data(pmbe->entry);
 
-	return flags;
-}
+	jump_to_uncached();
 
-/*
- * Must be run uncached.
- */
-static void __set_pmb_entry(struct pmb_entry *pmbe)
-{
-	writel_uncached(pmbe->vpn | PMB_V, mk_pmb_addr(pmbe->entry));
-	writel_uncached(pmbe->ppn | pmbe->flags | PMB_V,
-			mk_pmb_data(pmbe->entry));
+	/* Set V-bit */
+	__raw_writel(pmbe->vpn | PMB_V, addr);
+	__raw_writel(pmbe->ppn | pmbe->flags | PMB_V, data);
+
+	back_to_cached();
 }
 
 static void __clear_pmb_entry(struct pmb_entry *pmbe)
@@ -194,144 +332,155 @@ static void set_pmb_entry(struct pmb_entry *pmbe)
 	spin_unlock_irqrestore(&pmbe->lock, flags);
 }
 
-static struct {
-	unsigned long size;
-	int flag;
-} pmb_sizes[] = {
-	{ .size	= SZ_512M, .flag = PMB_SZ_512M, },
-	{ .size = SZ_128M, .flag = PMB_SZ_128M, },
-	{ .size = SZ_64M,  .flag = PMB_SZ_64M,  },
-	{ .size = SZ_16M,  .flag = PMB_SZ_16M,  },
-};
-
-long pmb_remap(unsigned long vaddr, unsigned long phys,
-	       unsigned long size, pgprot_t prot)
+int pmb_bolt_mapping(unsigned long vaddr, phys_addr_t phys,
+		     unsigned long size, pgprot_t prot)
 {
 	struct pmb_entry *pmbp, *pmbe;
-	unsigned long wanted;
-	int pmb_flags, i;
-	long err;
-	u64 flags;
+	unsigned long orig_addr, orig_size;
+	unsigned long flags, pmb_flags;
+	int i, mapped;
 
-	flags = pgprot_val(prot);
+	if (!pmb_addr_valid(vaddr, size))
+		return -EFAULT;
+	if (pmb_mapping_exists(vaddr, phys, size))
+		return 0;
 
-	pmb_flags = PMB_WT | PMB_UB;
-
-	/* Convert typical pgprot value to the PMB equivalent */
-	if (flags & _PAGE_CACHABLE) {
-		pmb_flags |= PMB_C;
+	orig_addr = vaddr;
+	orig_size = size;
 
-		if ((flags & _PAGE_WT) == 0)
-			pmb_flags &= ~(PMB_WT | PMB_UB);
-	}
+	flush_tlb_kernel_range(vaddr, vaddr + size);
 
+	pmb_flags = pgprot_to_pmb_flags(prot);
 	pmbp = NULL;
-	wanted = size;
 
-again:
-	for (i = 0; i < ARRAY_SIZE(pmb_sizes); i++) {
-		unsigned long flags;
+	do {
+		for (i = mapped = 0; i < ARRAY_SIZE(pmb_sizes); i++) {
+			if (size < pmb_sizes[i].size)
+				continue;
+
+			pmbe = pmb_alloc(vaddr, phys, pmb_flags |
+					 pmb_sizes[i].flag, PMB_NO_ENTRY);
+			if (IS_ERR(pmbe)) {
+				pmb_unmap_entry(pmbp, mapped);
+				return PTR_ERR(pmbe);
+			}
 
-		if (size < pmb_sizes[i].size)
-			continue;
+			spin_lock_irqsave(&pmbe->lock, flags);
 
-		pmbe = pmb_alloc(vaddr, phys, pmb_flags | pmb_sizes[i].flag,
-				 PMB_NO_ENTRY);
-		if (IS_ERR(pmbe)) {
-			err = PTR_ERR(pmbe);
-			goto out;
-		}
+			pmbe->size = pmb_sizes[i].size;
 
-		spin_lock_irqsave(&pmbe->lock, flags);
+			__set_pmb_entry(pmbe);
 
-		__set_pmb_entry(pmbe);
+			phys	+= pmbe->size;
+			vaddr	+= pmbe->size;
+			size	-= pmbe->size;
 
-		phys	+= pmb_sizes[i].size;
-		vaddr	+= pmb_sizes[i].size;
-		size	-= pmb_sizes[i].size;
+			/*
+			 * Link adjacent entries that span multiple PMB
+			 * entries for easier tear-down.
+			 */
+			if (likely(pmbp)) {
+				spin_lock(&pmbp->lock);
+				pmbp->link = pmbe;
+				spin_unlock(&pmbp->lock);
+			}
 
-		pmbe->size = pmb_sizes[i].size;
+			pmbp = pmbe;
 
-		/*
-		 * Link adjacent entries that span multiple PMB entries
-		 * for easier tear-down.
-		 */
-		if (likely(pmbp)) {
-			spin_lock(&pmbp->lock);
-			pmbp->link = pmbe;
-			spin_unlock(&pmbp->lock);
+			/*
+			 * Instead of trying smaller sizes on every
+			 * iteration (even if we succeed in allocating
+			 * space), try using pmb_sizes[i].size again.
+			 */
+			i--;
+			mapped++;
+
+			spin_unlock_irqrestore(&pmbe->lock, flags);
 		}
+	} while (size >= SZ_16M);
 
-		pmbp = pmbe;
+	flush_cache_vmap(orig_addr, orig_addr + orig_size);
 
-		/*
-		 * Instead of trying smaller sizes on every iteration
-		 * (even if we succeed in allocating space), try using
-		 * pmb_sizes[i].size again.
-		 */
-		i--;
+	return 0;
+}
 
-		spin_unlock_irqrestore(&pmbe->lock, flags);
-	}
+void __iomem *pmb_remap_caller(phys_addr_t phys, unsigned long size,
+			       pgprot_t prot, void *caller)
+{
+	unsigned long vaddr;
+	phys_addr_t offset, last_addr;
+	phys_addr_t align_mask;
+	unsigned long aligned;
+	struct vm_struct *area;
+	int i, ret;
 
-	if (size >= SZ_16M)
-		goto again;
+	if (!pmb_iomapping_enabled)
+		return NULL;
 
-	return wanted - size;
+	/*
+	 * Small mappings need to go through the TLB.
+	 */
+	if (size < SZ_16M)
+		return ERR_PTR(-EINVAL);
+	if (!pmb_prot_valid(prot))
+		return ERR_PTR(-EINVAL);
 
-out:
-	pmb_unmap_entry(pmbp, NR_PMB_ENTRIES);
+	for (i = 0; i < ARRAY_SIZE(pmb_sizes); i++)
+		if (size >= pmb_sizes[i].size)
+			break;
+
+	last_addr = phys + size;
+	align_mask = ~(pmb_sizes[i].size - 1);
+	offset = phys & ~align_mask;
+	phys &= align_mask;
+	aligned = ALIGN(last_addr, pmb_sizes[i].size) - phys;
+
+	/*
+	 * XXX: This should really start from uncached_end, but this
+	 * causes the MMU to reset, so for now we restrict it to the
+	 * 0xb000...0xc000 range.
+	 */
+	area = __get_vm_area_caller(aligned, VM_IOREMAP, 0xb0000000,
+				    P3SEG, caller);
+	if (!area)
+		return NULL;
+
+	area->phys_addr = phys;
+	vaddr = (unsigned long)area->addr;
+
+	ret = pmb_bolt_mapping(vaddr, phys, size, prot);
+	if (unlikely(ret != 0))
+		return ERR_PTR(ret);
 
-	return err;
+	return (void __iomem *)(offset + (char *)vaddr);
 }
 
-void pmb_unmap(unsigned long addr)
+int pmb_unmap(void __iomem *addr)
 {
 	struct pmb_entry *pmbe = NULL;
-	int i;
+	unsigned long vaddr = (unsigned long __force)addr;
+	int i, found = 0;
 
 	read_lock(&pmb_rwlock);
 
 	for (i = 0; i < ARRAY_SIZE(pmb_entry_list); i++) {
 		if (test_bit(i, pmb_map)) {
 			pmbe = &pmb_entry_list[i];
-			if (pmbe->vpn == addr)
+			if (pmbe->vpn == vaddr) {
+				found = 1;
 				break;
+			}
 		}
 	}
 
 	read_unlock(&pmb_rwlock);
 
-	pmb_unmap_entry(pmbe, NR_PMB_ENTRIES);
-}
-
-static bool pmb_can_merge(struct pmb_entry *a, struct pmb_entry *b)
-{
-	return (b->vpn == (a->vpn + a->size)) &&
-	       (b->ppn == (a->ppn + a->size)) &&
-	       (b->flags == a->flags);
-}
-
-static bool pmb_size_valid(unsigned long size)
-{
-	int i;
-
-	for (i = 0; i < ARRAY_SIZE(pmb_sizes); i++)
-		if (pmb_sizes[i].size == size)
-			return true;
-
-	return false;
-}
-
-static int pmb_size_to_flags(unsigned long size)
-{
-	int i;
-
-	for (i = 0; i < ARRAY_SIZE(pmb_sizes); i++)
-		if (pmb_sizes[i].size == size)
-			return pmb_sizes[i].flag;
+	if (found) {
+		pmb_unmap_entry(pmbe, NR_PMB_ENTRIES);
+		return 0;
+	}
 
-	return 0;
+	return -EINVAL;
 }
 
 static void __pmb_unmap_entry(struct pmb_entry *pmbe, int depth)
@@ -351,6 +500,8 @@ static void __pmb_unmap_entry(struct pmb_entry *pmbe, int depth)
 		 */
 		__clear_pmb_entry(pmbe);
 
+		flush_cache_vunmap(pmbe->vpn, pmbe->vpn + pmbe->size);
+
 		pmbe = pmblink->link;
 
 		pmb_free(pmblink);
@@ -369,11 +520,6 @@ static void pmb_unmap_entry(struct pmb_entry *pmbe, int depth)
 	write_unlock_irqrestore(&pmb_rwlock, flags);
 }
 
-static __always_inline unsigned int pmb_ppn_in_range(unsigned long ppn)
-{
-	return ppn >= __pa(memory_start) && ppn < __pa(memory_end);
-}
-
 static void __init pmb_notify(void)
 {
 	int i;
@@ -625,6 +771,18 @@ static void __init pmb_resize(void)
 }
 #endif
 
+static int __init early_pmb(char *p)
+{
+	if (!p)
+		return 0;
+
+	if (strstr(p, "iomap"))
+		pmb_iomapping_enabled = 1;
+
+	return 0;
+}
+early_param("pmb", early_pmb);
+
 void __init pmb_init(void)
 {
 	/* Synchronize software state */
@@ -713,7 +871,7 @@ static int __init pmb_debugfs_init(void)
 
 	return 0;
 }
-postcore_initcall(pmb_debugfs_init);
+subsys_initcall(pmb_debugfs_init);
 
 #ifdef CONFIG_PM
 static int pmb_sysdev_suspend(struct sys_device *dev, pm_message_t state)
diff --git a/drivers/dma/shdma.c b/drivers/dma/shdma.c
index b75ce8b84c46..5d17e09cb625 100644
--- a/drivers/dma/shdma.c
+++ b/drivers/dma/shdma.c
@@ -24,8 +24,10 @@
 #include <linux/delay.h>
 #include <linux/dma-mapping.h>
 #include <linux/platform_device.h>
-#include <cpu/dma.h>
-#include <asm/dma-sh.h>
+#include <linux/pm_runtime.h>
+
+#include <asm/dmaengine.h>
+
 #include "shdma.h"
 
 /* DMA descriptor control */
@@ -38,30 +40,32 @@ enum sh_dmae_desc_status {
 };
 
 #define NR_DESCS_PER_CHANNEL 32
-/*
- * Define the default configuration for dual address memory-memory transfer.
- * The 0x400 value represents auto-request, external->external.
- *
- * And this driver set 4byte burst mode.
- * If you want to change mode, you need to change RS_DEFAULT of value.
- * (ex 1byte burst mode -> (RS_DUAL & ~TS_32)
- */
-#define RS_DEFAULT  (RS_DUAL)
+/* Default MEMCPY transfer size = 2^2 = 4 bytes */
+#define LOG2_DEFAULT_XFER_SIZE	2
 
 /* A bitmask with bits enough for enum sh_dmae_slave_chan_id */
 static unsigned long sh_dmae_slave_used[BITS_TO_LONGS(SHDMA_SLAVE_NUMBER)];
 
 static void sh_dmae_chan_ld_cleanup(struct sh_dmae_chan *sh_chan, bool all);
 
-#define SH_DMAC_CHAN_BASE(id) (dma_base_addr[id])
 static void sh_dmae_writel(struct sh_dmae_chan *sh_dc, u32 data, u32 reg)
 {
-	ctrl_outl(data, SH_DMAC_CHAN_BASE(sh_dc->id) + reg);
+	__raw_writel(data, sh_dc->base + reg / sizeof(u32));
 }
 
 static u32 sh_dmae_readl(struct sh_dmae_chan *sh_dc, u32 reg)
 {
-	return ctrl_inl(SH_DMAC_CHAN_BASE(sh_dc->id) + reg);
+	return __raw_readl(sh_dc->base + reg / sizeof(u32));
+}
+
+static u16 dmaor_read(struct sh_dmae_device *shdev)
+{
+	return __raw_readw(shdev->chan_reg + DMAOR / sizeof(u32));
+}
+
+static void dmaor_write(struct sh_dmae_device *shdev, u16 data)
+{
+	__raw_writew(data, shdev->chan_reg + DMAOR / sizeof(u32));
 }
 
 /*
@@ -69,24 +73,23 @@ static u32 sh_dmae_readl(struct sh_dmae_chan *sh_dc, u32 reg)
  *
  * SH7780 has two DMAOR register
  */
-static void sh_dmae_ctl_stop(int id)
+static void sh_dmae_ctl_stop(struct sh_dmae_device *shdev)
 {
-	unsigned short dmaor = dmaor_read_reg(id);
+	unsigned short dmaor = dmaor_read(shdev);
 
-	dmaor &= ~(DMAOR_NMIF | DMAOR_AE);
-	dmaor_write_reg(id, dmaor);
+	dmaor_write(shdev, dmaor & ~(DMAOR_NMIF | DMAOR_AE | DMAOR_DME));
 }
 
-static int sh_dmae_rst(int id)
+static int sh_dmae_rst(struct sh_dmae_device *shdev)
 {
 	unsigned short dmaor;
 
-	sh_dmae_ctl_stop(id);
-	dmaor = dmaor_read_reg(id) | DMAOR_INIT;
+	sh_dmae_ctl_stop(shdev);
+	dmaor = dmaor_read(shdev) | shdev->pdata->dmaor_init;
 
-	dmaor_write_reg(id, dmaor);
-	if (dmaor_read_reg(id) & (DMAOR_AE | DMAOR_NMIF)) {
-		pr_warning(KERN_ERR "dma-sh: Can't initialize DMAOR.\n");
+	dmaor_write(shdev, dmaor);
+	if (dmaor_read(shdev) & (DMAOR_AE | DMAOR_NMIF)) {
+		pr_warning("dma-sh: Can't initialize DMAOR.\n");
 		return -EINVAL;
 	}
 	return 0;
@@ -102,13 +105,36 @@ static bool dmae_is_busy(struct sh_dmae_chan *sh_chan)
 	return false; /* waiting */
 }
 
-static unsigned int ts_shift[] = TS_SHIFT;
-static inline unsigned int calc_xmit_shift(u32 chcr)
+static unsigned int calc_xmit_shift(struct sh_dmae_chan *sh_chan, u32 chcr)
 {
-	int cnt = ((chcr & CHCR_TS_LOW_MASK) >> CHCR_TS_LOW_SHIFT) |
-		((chcr & CHCR_TS_HIGH_MASK) >> CHCR_TS_HIGH_SHIFT);
+	struct sh_dmae_device *shdev = container_of(sh_chan->common.device,
+						struct sh_dmae_device, common);
+	struct sh_dmae_pdata *pdata = shdev->pdata;
+	int cnt = ((chcr & pdata->ts_low_mask) >> pdata->ts_low_shift) |
+		((chcr & pdata->ts_high_mask) >> pdata->ts_high_shift);
+
+	if (cnt >= pdata->ts_shift_num)
+		cnt = 0;
 
-	return ts_shift[cnt];
+	return pdata->ts_shift[cnt];
+}
+
+static u32 log2size_to_chcr(struct sh_dmae_chan *sh_chan, int l2size)
+{
+	struct sh_dmae_device *shdev = container_of(sh_chan->common.device,
+						struct sh_dmae_device, common);
+	struct sh_dmae_pdata *pdata = shdev->pdata;
+	int i;
+
+	for (i = 0; i < pdata->ts_shift_num; i++)
+		if (pdata->ts_shift[i] == l2size)
+			break;
+
+	if (i == pdata->ts_shift_num)
+		i = 0;
+
+	return ((i << pdata->ts_low_shift) & pdata->ts_low_mask) |
+		((i << pdata->ts_high_shift) & pdata->ts_high_mask);
 }
 
 static void dmae_set_reg(struct sh_dmae_chan *sh_chan, struct sh_dmae_regs *hw)
@@ -136,8 +162,13 @@ static void dmae_halt(struct sh_dmae_chan *sh_chan)
 
 static void dmae_init(struct sh_dmae_chan *sh_chan)
 {
-	u32 chcr = RS_DEFAULT; /* default is DUAL mode */
-	sh_chan->xmit_shift = calc_xmit_shift(chcr);
+	/*
+	 * Default configuration for dual address memory-memory transfer.
+	 * 0x400 represents auto-request.
+	 */
+	u32 chcr = DM_INC | SM_INC | 0x400 | log2size_to_chcr(sh_chan,
+						   LOG2_DEFAULT_XFER_SIZE);
+	sh_chan->xmit_shift = calc_xmit_shift(sh_chan, chcr);
 	sh_dmae_writel(sh_chan, chcr, CHCR);
 }
 
@@ -147,37 +178,26 @@ static int dmae_set_chcr(struct sh_dmae_chan *sh_chan, u32 val)
 	if (dmae_is_busy(sh_chan))
 		return -EBUSY;
 
-	sh_chan->xmit_shift = calc_xmit_shift(val);
+	sh_chan->xmit_shift = calc_xmit_shift(sh_chan, val);
 	sh_dmae_writel(sh_chan, val, CHCR);
 
 	return 0;
 }
 
-#define DMARS_SHIFT	8
-#define DMARS_CHAN_MSK	0x01
 static int dmae_set_dmars(struct sh_dmae_chan *sh_chan, u16 val)
 {
-	u32 addr;
-	int shift = 0;
+	struct sh_dmae_device *shdev = container_of(sh_chan->common.device,
+						struct sh_dmae_device, common);
+	struct sh_dmae_pdata *pdata = shdev->pdata;
+	struct sh_dmae_channel *chan_pdata = &pdata->channel[sh_chan->id];
+	u16 __iomem *addr = shdev->dmars + chan_pdata->dmars / sizeof(u16);
+	int shift = chan_pdata->dmars_bit;
 
 	if (dmae_is_busy(sh_chan))
 		return -EBUSY;
 
-	if (sh_chan->id & DMARS_CHAN_MSK)
-		shift = DMARS_SHIFT;
-
-	if (sh_chan->id < 6)
-		/* DMA0RS0 - DMA0RS2 */
-		addr = SH_DMARS_BASE0 + (sh_chan->id / 2) * 4;
-#ifdef SH_DMARS_BASE1
-	else if (sh_chan->id < 12)
-		/* DMA1RS0 - DMA1RS2 */
-		addr = SH_DMARS_BASE1 + ((sh_chan->id - 6) / 2) * 4;
-#endif
-	else
-		return -EINVAL;
-
-	ctrl_outw((val << shift) | (ctrl_inw(addr) & (0xFF00 >> shift)), addr);
+	__raw_writew((__raw_readw(addr) & (0xff00 >> shift)) | (val << shift),
+		     addr);
 
 	return 0;
 }
@@ -251,15 +271,15 @@ static struct sh_dmae_slave_config *sh_dmae_find_slave(
 	struct dma_device *dma_dev = sh_chan->common.device;
 	struct sh_dmae_device *shdev = container_of(dma_dev,
 					struct sh_dmae_device, common);
-	struct sh_dmae_pdata *pdata = &shdev->pdata;
+	struct sh_dmae_pdata *pdata = shdev->pdata;
 	int i;
 
 	if ((unsigned)slave_id >= SHDMA_SLAVE_NUMBER)
 		return NULL;
 
-	for (i = 0; i < pdata->config_num; i++)
-		if (pdata->config[i].slave_id == slave_id)
-			return pdata->config + i;
+	for (i = 0; i < pdata->slave_num; i++)
+		if (pdata->slave[i].slave_id == slave_id)
+			return pdata->slave + i;
 
 	return NULL;
 }
@@ -270,6 +290,8 @@ static int sh_dmae_alloc_chan_resources(struct dma_chan *chan)
 	struct sh_desc *desc;
 	struct sh_dmae_slave *param = chan->private;
 
+	pm_runtime_get_sync(sh_chan->dev);
+
 	/*
 	 * This relies on the guarantee from dmaengine that alloc_chan_resources
 	 * never runs concurrently with itself or free_chan_resources.
@@ -288,9 +310,8 @@ static int sh_dmae_alloc_chan_resources(struct dma_chan *chan)
 
 		dmae_set_dmars(sh_chan, cfg->mid_rid);
 		dmae_set_chcr(sh_chan, cfg->chcr);
-	} else {
-		if ((sh_dmae_readl(sh_chan, CHCR) & 0x700) != 0x400)
-			dmae_set_chcr(sh_chan, RS_DEFAULT);
+	} else if ((sh_dmae_readl(sh_chan, CHCR) & 0xf00) != 0x400) {
+		dmae_init(sh_chan);
 	}
 
 	spin_lock_bh(&sh_chan->desc_lock);
@@ -312,6 +333,9 @@ static int sh_dmae_alloc_chan_resources(struct dma_chan *chan)
 	}
 	spin_unlock_bh(&sh_chan->desc_lock);
 
+	if (!sh_chan->descs_allocated)
+		pm_runtime_put(sh_chan->dev);
+
 	return sh_chan->descs_allocated;
 }
 
@@ -323,6 +347,7 @@ static void sh_dmae_free_chan_resources(struct dma_chan *chan)
 	struct sh_dmae_chan *sh_chan = to_sh_chan(chan);
 	struct sh_desc *desc, *_desc;
 	LIST_HEAD(list);
+	int descs = sh_chan->descs_allocated;
 
 	dmae_halt(sh_chan);
 
@@ -343,6 +368,9 @@ static void sh_dmae_free_chan_resources(struct dma_chan *chan)
 
 	spin_unlock_bh(&sh_chan->desc_lock);
 
+	if (descs > 0)
+		pm_runtime_put(sh_chan->dev);
+
 	list_for_each_entry_safe(desc, _desc, &list, node)
 		kfree(desc);
 }
@@ -559,6 +587,19 @@ static void sh_dmae_terminate_all(struct dma_chan *chan)
 	if (!chan)
 		return;
 
+	dmae_halt(sh_chan);
+
+	spin_lock_bh(&sh_chan->desc_lock);
+	if (!list_empty(&sh_chan->ld_queue)) {
+		/* Record partial transfer */
+		struct sh_desc *desc = list_entry(sh_chan->ld_queue.next,
+						  struct sh_desc, node);
+		desc->partial = (desc->hw.tcr - sh_dmae_readl(sh_chan, TCR)) <<
+			sh_chan->xmit_shift;
+
+	}
+	spin_unlock_bh(&sh_chan->desc_lock);
+
 	sh_dmae_chan_ld_cleanup(sh_chan, true);
 }
 
@@ -661,7 +702,7 @@ static void sh_dmae_chan_ld_cleanup(struct sh_dmae_chan *sh_chan, bool all)
 
 static void sh_chan_xfer_ld_queue(struct sh_dmae_chan *sh_chan)
 {
-	struct sh_desc *sd;
+	struct sh_desc *desc;
 
 	spin_lock_bh(&sh_chan->desc_lock);
 	/* DMA work check */
@@ -671,10 +712,13 @@ static void sh_chan_xfer_ld_queue(struct sh_dmae_chan *sh_chan)
 	}
 
 	/* Find the first not transferred desciptor */
-	list_for_each_entry(sd, &sh_chan->ld_queue, node)
-		if (sd->mark == DESC_SUBMITTED) {
+	list_for_each_entry(desc, &sh_chan->ld_queue, node)
+		if (desc->mark == DESC_SUBMITTED) {
+			dev_dbg(sh_chan->dev, "Queue #%d to %d: %u@%x -> %x\n",
+				desc->async_tx.cookie, sh_chan->id,
+				desc->hw.tcr, desc->hw.sar, desc->hw.dar);
 			/* Get the ld start address from ld_queue */
-			dmae_set_reg(sh_chan, &sd->hw);
+			dmae_set_reg(sh_chan, &desc->hw);
 			dmae_start(sh_chan);
 			break;
 		}
@@ -696,6 +740,7 @@ static enum dma_status sh_dmae_is_complete(struct dma_chan *chan,
 	struct sh_dmae_chan *sh_chan = to_sh_chan(chan);
 	dma_cookie_t last_used;
 	dma_cookie_t last_complete;
+	enum dma_status status;
 
 	sh_dmae_chan_ld_cleanup(sh_chan, false);
 
@@ -709,7 +754,27 @@ static enum dma_status sh_dmae_is_complete(struct dma_chan *chan,
 	if (used)
 		*used = last_used;
 
-	return dma_async_is_complete(cookie, last_complete, last_used);
+	spin_lock_bh(&sh_chan->desc_lock);
+
+	status = dma_async_is_complete(cookie, last_complete, last_used);
+
+	/*
+	 * If we don't find cookie on the queue, it has been aborted and we have
+	 * to report error
+	 */
+	if (status != DMA_SUCCESS) {
+		struct sh_desc *desc;
+		status = DMA_ERROR;
+		list_for_each_entry(desc, &sh_chan->ld_queue, node)
+			if (desc->cookie == cookie) {
+				status = DMA_IN_PROGRESS;
+				break;
+			}
+	}
+
+	spin_unlock_bh(&sh_chan->desc_lock);
+
+	return status;
 }
 
 static irqreturn_t sh_dmae_interrupt(int irq, void *data)
@@ -732,40 +797,32 @@ static irqreturn_t sh_dmae_interrupt(int irq, void *data)
 #if defined(CONFIG_CPU_SH4)
 static irqreturn_t sh_dmae_err(int irq, void *data)
 {
-	int err = 0;
 	struct sh_dmae_device *shdev = (struct sh_dmae_device *)data;
+	int i;
 
-	/* IRQ Multi */
-	if (shdev->pdata.mode & SHDMA_MIX_IRQ) {
-		int __maybe_unused cnt = 0;
-		switch (irq) {
-#if defined(DMTE6_IRQ) && defined(DMAE1_IRQ)
-		case DMTE6_IRQ:
-			cnt++;
-#endif
-		case DMTE0_IRQ:
-			if (dmaor_read_reg(cnt) & (DMAOR_NMIF | DMAOR_AE)) {
-				disable_irq(irq);
-				return IRQ_HANDLED;
+	/* halt the dma controller */
+	sh_dmae_ctl_stop(shdev);
+
+	/* We cannot detect, which channel caused the error, have to reset all */
+	for (i = 0; i < SH_DMAC_MAX_CHANNELS; i++) {
+		struct sh_dmae_chan *sh_chan = shdev->chan[i];
+		if (sh_chan) {
+			struct sh_desc *desc;
+			/* Stop the channel */
+			dmae_halt(sh_chan);
+			/* Complete all  */
+			list_for_each_entry(desc, &sh_chan->ld_queue, node) {
+				struct dma_async_tx_descriptor *tx = &desc->async_tx;
+				desc->mark = DESC_IDLE;
+				if (tx->callback)
+					tx->callback(tx->callback_param);
 			}
-		default:
-			return IRQ_NONE;
+			list_splice_init(&sh_chan->ld_queue, &sh_chan->ld_free);
 		}
-	} else {
-		/* reset dma controller */
-		err = sh_dmae_rst(0);
-		if (err)
-			return err;
-#ifdef SH_DMAC_BASE1
-		if (shdev->pdata.mode & SHDMA_DMAOR1) {
-			err = sh_dmae_rst(1);
-			if (err)
-				return err;
-		}
-#endif
-		disable_irq(irq);
-		return IRQ_HANDLED;
 	}
+	sh_dmae_rst(shdev);
+
+	return IRQ_HANDLED;
 }
 #endif
 
@@ -796,19 +853,12 @@ static void dmae_do_tasklet(unsigned long data)
 	sh_dmae_chan_ld_cleanup(sh_chan, false);
 }
 
-static unsigned int get_dmae_irq(unsigned int id)
-{
-	unsigned int irq = 0;
-	if (id < ARRAY_SIZE(dmte_irq_map))
-		irq = dmte_irq_map[id];
-	return irq;
-}
-
-static int __devinit sh_dmae_chan_probe(struct sh_dmae_device *shdev, int id)
+static int __devinit sh_dmae_chan_probe(struct sh_dmae_device *shdev, int id,
+					int irq, unsigned long flags)
 {
 	int err;
-	unsigned int irq = get_dmae_irq(id);
-	unsigned long irqflags = IRQF_DISABLED;
+	struct sh_dmae_channel *chan_pdata = &shdev->pdata->channel[id];
+	struct platform_device *pdev = to_platform_device(shdev->common.dev);
 	struct sh_dmae_chan *new_sh_chan;
 
 	/* alloc channel */
@@ -819,8 +869,13 @@ static int __devinit sh_dmae_chan_probe(struct sh_dmae_device *shdev, int id)
 		return -ENOMEM;
 	}
 
+	/* copy struct dma_device */
+	new_sh_chan->common.device = &shdev->common;
+
 	new_sh_chan->dev = shdev->common.dev;
 	new_sh_chan->id = id;
+	new_sh_chan->irq = irq;
+	new_sh_chan->base = shdev->chan_reg + chan_pdata->offset / sizeof(u32);
 
 	/* Init DMA tasklet */
 	tasklet_init(&new_sh_chan->tasklet, dmae_do_tasklet,
@@ -835,29 +890,20 @@ static int __devinit sh_dmae_chan_probe(struct sh_dmae_device *shdev, int id)
 	INIT_LIST_HEAD(&new_sh_chan->ld_queue);
 	INIT_LIST_HEAD(&new_sh_chan->ld_free);
 
-	/* copy struct dma_device */
-	new_sh_chan->common.device = &shdev->common;
-
 	/* Add the channel to DMA device channel list */
 	list_add_tail(&new_sh_chan->common.device_node,
 			&shdev->common.channels);
 	shdev->common.chancnt++;
 
-	if (shdev->pdata.mode & SHDMA_MIX_IRQ) {
-		irqflags = IRQF_SHARED;
-#if defined(DMTE6_IRQ)
-		if (irq >= DMTE6_IRQ)
-			irq = DMTE6_IRQ;
-		else
-#endif
-			irq = DMTE0_IRQ;
-	}
-
-	snprintf(new_sh_chan->dev_id, sizeof(new_sh_chan->dev_id),
-		 "sh-dmae%d", new_sh_chan->id);
+	if (pdev->id >= 0)
+		snprintf(new_sh_chan->dev_id, sizeof(new_sh_chan->dev_id),
+			 "sh-dmae%d.%d", pdev->id, new_sh_chan->id);
+	else
+		snprintf(new_sh_chan->dev_id, sizeof(new_sh_chan->dev_id),
+			 "sh-dma%d", new_sh_chan->id);
 
 	/* set up channel irq */
-	err = request_irq(irq, &sh_dmae_interrupt, irqflags,
+	err = request_irq(irq, &sh_dmae_interrupt, flags,
 			  new_sh_chan->dev_id, new_sh_chan);
 	if (err) {
 		dev_err(shdev->common.dev, "DMA channel %d request_irq error "
@@ -881,12 +927,12 @@ static void sh_dmae_chan_remove(struct sh_dmae_device *shdev)
 
 	for (i = shdev->common.chancnt - 1 ; i >= 0 ; i--) {
 		if (shdev->chan[i]) {
-			struct sh_dmae_chan *shchan = shdev->chan[i];
-			if (!(shdev->pdata.mode & SHDMA_MIX_IRQ))
-				free_irq(dmte_irq_map[i], shchan);
+			struct sh_dmae_chan *sh_chan = shdev->chan[i];
 
-			list_del(&shchan->common.device_node);
-			kfree(shchan);
+			free_irq(sh_chan->irq, sh_chan);
+
+			list_del(&sh_chan->common.device_node);
+			kfree(sh_chan);
 			shdev->chan[i] = NULL;
 		}
 	}
@@ -895,47 +941,84 @@ static void sh_dmae_chan_remove(struct sh_dmae_device *shdev)
 
 static int __init sh_dmae_probe(struct platform_device *pdev)
 {
-	int err = 0, cnt, ecnt;
-	unsigned long irqflags = IRQF_DISABLED;
-#if defined(CONFIG_CPU_SH4)
-	int eirq[] = { DMAE0_IRQ,
-#if defined(DMAE1_IRQ)
-			DMAE1_IRQ
-#endif
-		};
-#endif
+	struct sh_dmae_pdata *pdata = pdev->dev.platform_data;
+	unsigned long irqflags = IRQF_DISABLED,
+		chan_flag[SH_DMAC_MAX_CHANNELS] = {};
+	int errirq, chan_irq[SH_DMAC_MAX_CHANNELS];
+	int err, i, irq_cnt = 0, irqres = 0;
 	struct sh_dmae_device *shdev;
+	struct resource *chan, *dmars, *errirq_res, *chanirq_res;
 
 	/* get platform data */
-	if (!pdev->dev.platform_data)
+	if (!pdata || !pdata->channel_num)
 		return -ENODEV;
 
+	chan = platform_get_resource(pdev, IORESOURCE_MEM, 0);
+	/* DMARS area is optional, if absent, this controller cannot do slave DMA */
+	dmars = platform_get_resource(pdev, IORESOURCE_MEM, 1);
+	/*
+	 * IRQ resources:
+	 * 1. there always must be at least one IRQ IO-resource. On SH4 it is
+	 *    the error IRQ, in which case it is the only IRQ in this resource:
+	 *    start == end. If it is the only IRQ resource, all channels also
+	 *    use the same IRQ.
+	 * 2. DMA channel IRQ resources can be specified one per resource or in
+	 *    ranges (start != end)
+	 * 3. iff all events (channels and, optionally, error) on this
+	 *    controller use the same IRQ, only one IRQ resource can be
+	 *    specified, otherwise there must be one IRQ per channel, even if
+	 *    some of them are equal
+	 * 4. if all IRQs on this controller are equal or if some specific IRQs
+	 *    specify IORESOURCE_IRQ_SHAREABLE in their resources, they will be
+	 *    requested with the IRQF_SHARED flag
+	 */
+	errirq_res = platform_get_resource(pdev, IORESOURCE_IRQ, 0);
+	if (!chan || !errirq_res)
+		return -ENODEV;
+
+	if (!request_mem_region(chan->start, resource_size(chan), pdev->name)) {
+		dev_err(&pdev->dev, "DMAC register region already claimed\n");
+		return -EBUSY;
+	}
+
+	if (dmars && !request_mem_region(dmars->start, resource_size(dmars), pdev->name)) {
+		dev_err(&pdev->dev, "DMAC DMARS region already claimed\n");
+		err = -EBUSY;
+		goto ermrdmars;
+	}
+
+	err = -ENOMEM;
 	shdev = kzalloc(sizeof(struct sh_dmae_device), GFP_KERNEL);
 	if (!shdev) {
-		dev_err(&pdev->dev, "No enough memory\n");
-		return -ENOMEM;
+		dev_err(&pdev->dev, "Not enough memory\n");
+		goto ealloc;
+	}
+
+	shdev->chan_reg = ioremap(chan->start, resource_size(chan));
+	if (!shdev->chan_reg)
+		goto emapchan;
+	if (dmars) {
+		shdev->dmars = ioremap(dmars->start, resource_size(dmars));
+		if (!shdev->dmars)
+			goto emapdmars;
 	}
 
 	/* platform data */
-	memcpy(&shdev->pdata, pdev->dev.platform_data,
-			sizeof(struct sh_dmae_pdata));
+	shdev->pdata = pdata;
+
+	pm_runtime_enable(&pdev->dev);
+	pm_runtime_get_sync(&pdev->dev);
 
 	/* reset dma controller */
-	err = sh_dmae_rst(0);
+	err = sh_dmae_rst(shdev);
 	if (err)
 		goto rst_err;
 
-	/* SH7780/85/23 has DMAOR1 */
-	if (shdev->pdata.mode & SHDMA_DMAOR1) {
-		err = sh_dmae_rst(1);
-		if (err)
-			goto rst_err;
-	}
-
 	INIT_LIST_HEAD(&shdev->common.channels);
 
 	dma_cap_set(DMA_MEMCPY, shdev->common.cap_mask);
-	dma_cap_set(DMA_SLAVE, shdev->common.cap_mask);
+	if (dmars)
+		dma_cap_set(DMA_SLAVE, shdev->common.cap_mask);
 
 	shdev->common.device_alloc_chan_resources
 		= sh_dmae_alloc_chan_resources;
@@ -950,37 +1033,72 @@ static int __init sh_dmae_probe(struct platform_device *pdev)
 
 	shdev->common.dev = &pdev->dev;
 	/* Default transfer size of 32 bytes requires 32-byte alignment */
-	shdev->common.copy_align = 5;
+	shdev->common.copy_align = LOG2_DEFAULT_XFER_SIZE;
 
 #if defined(CONFIG_CPU_SH4)
-	/* Non Mix IRQ mode SH7722/SH7730 etc... */
-	if (shdev->pdata.mode & SHDMA_MIX_IRQ) {
+	chanirq_res = platform_get_resource(pdev, IORESOURCE_IRQ, 1);
+
+	if (!chanirq_res)
+		chanirq_res = errirq_res;
+	else
+		irqres++;
+
+	if (chanirq_res == errirq_res ||
+	    (errirq_res->flags & IORESOURCE_BITS) == IORESOURCE_IRQ_SHAREABLE)
 		irqflags = IRQF_SHARED;
-		eirq[0] = DMTE0_IRQ;
-#if defined(DMTE6_IRQ) && defined(DMAE1_IRQ)
-		eirq[1] = DMTE6_IRQ;
-#endif
+
+	errirq = errirq_res->start;
+
+	err = request_irq(errirq, sh_dmae_err, irqflags,
+			  "DMAC Address Error", shdev);
+	if (err) {
+		dev_err(&pdev->dev,
+			"DMA failed requesting irq #%d, error %d\n",
+			errirq, err);
+		goto eirq_err;
 	}
 
-	for (ecnt = 0 ; ecnt < ARRAY_SIZE(eirq); ecnt++) {
-		err = request_irq(eirq[ecnt], sh_dmae_err, irqflags,
-				  "DMAC Address Error", shdev);
-		if (err) {
-			dev_err(&pdev->dev, "DMA device request_irq"
-				"error (irq %d) with return %d\n",
-				eirq[ecnt], err);
-			goto eirq_err;
+#else
+	chanirq_res = errirq_res;
+#endif /* CONFIG_CPU_SH4 */
+
+	if (chanirq_res->start == chanirq_res->end &&
+	    !platform_get_resource(pdev, IORESOURCE_IRQ, 1)) {
+		/* Special case - all multiplexed */
+		for (; irq_cnt < pdata->channel_num; irq_cnt++) {
+			chan_irq[irq_cnt] = chanirq_res->start;
+			chan_flag[irq_cnt] = IRQF_SHARED;
 		}
+	} else {
+		do {
+			for (i = chanirq_res->start; i <= chanirq_res->end; i++) {
+				if ((errirq_res->flags & IORESOURCE_BITS) ==
+				    IORESOURCE_IRQ_SHAREABLE)
+					chan_flag[irq_cnt] = IRQF_SHARED;
+				else
+					chan_flag[irq_cnt] = IRQF_DISABLED;
+				dev_dbg(&pdev->dev,
+					"Found IRQ %d for channel %d\n",
+					i, irq_cnt);
+				chan_irq[irq_cnt++] = i;
+			}
+			chanirq_res = platform_get_resource(pdev,
+						IORESOURCE_IRQ, ++irqres);
+		} while (irq_cnt < pdata->channel_num && chanirq_res);
 	}
-#endif /* CONFIG_CPU_SH4 */
+
+	if (irq_cnt < pdata->channel_num)
+		goto eirqres;
 
 	/* Create DMA Channel */
-	for (cnt = 0 ; cnt < MAX_DMA_CHANNELS ; cnt++) {
-		err = sh_dmae_chan_probe(shdev, cnt);
+	for (i = 0; i < pdata->channel_num; i++) {
+		err = sh_dmae_chan_probe(shdev, i, chan_irq[i], chan_flag[i]);
 		if (err)
 			goto chan_probe_err;
 	}
 
+	pm_runtime_put(&pdev->dev);
+
 	platform_set_drvdata(pdev, shdev);
 	dma_async_device_register(&shdev->common);
 
@@ -988,13 +1106,24 @@ static int __init sh_dmae_probe(struct platform_device *pdev)
 
 chan_probe_err:
 	sh_dmae_chan_remove(shdev);
-
+eirqres:
+#if defined(CONFIG_CPU_SH4)
+	free_irq(errirq, shdev);
 eirq_err:
-	for (ecnt-- ; ecnt >= 0; ecnt--)
-		free_irq(eirq[ecnt], shdev);
-
+#endif
 rst_err:
+	pm_runtime_put(&pdev->dev);
+	if (dmars)
+		iounmap(shdev->dmars);
+emapdmars:
+	iounmap(shdev->chan_reg);
+emapchan:
 	kfree(shdev);
+ealloc:
+	if (dmars)
+		release_mem_region(dmars->start, resource_size(dmars));
+ermrdmars:
+	release_mem_region(chan->start, resource_size(chan));
 
 	return err;
 }
@@ -1002,36 +1131,39 @@ rst_err:
 static int __exit sh_dmae_remove(struct platform_device *pdev)
 {
 	struct sh_dmae_device *shdev = platform_get_drvdata(pdev);
+	struct resource *res;
+	int errirq = platform_get_irq(pdev, 0);
 
 	dma_async_device_unregister(&shdev->common);
 
-	if (shdev->pdata.mode & SHDMA_MIX_IRQ) {
-		free_irq(DMTE0_IRQ, shdev);
-#if defined(DMTE6_IRQ)
-		free_irq(DMTE6_IRQ, shdev);
-#endif
-	}
+	if (errirq > 0)
+		free_irq(errirq, shdev);
 
 	/* channel data remove */
 	sh_dmae_chan_remove(shdev);
 
-	if (!(shdev->pdata.mode & SHDMA_MIX_IRQ)) {
-		free_irq(DMAE0_IRQ, shdev);
-#if defined(DMAE1_IRQ)
-		free_irq(DMAE1_IRQ, shdev);
-#endif
-	}
+	pm_runtime_disable(&pdev->dev);
+
+	if (shdev->dmars)
+		iounmap(shdev->dmars);
+	iounmap(shdev->chan_reg);
+
 	kfree(shdev);
 
+	res = platform_get_resource(pdev, IORESOURCE_MEM, 0);
+	if (res)
+		release_mem_region(res->start, resource_size(res));
+	res = platform_get_resource(pdev, IORESOURCE_MEM, 1);
+	if (res)
+		release_mem_region(res->start, resource_size(res));
+
 	return 0;
 }
 
 static void sh_dmae_shutdown(struct platform_device *pdev)
 {
 	struct sh_dmae_device *shdev = platform_get_drvdata(pdev);
-	sh_dmae_ctl_stop(0);
-	if (shdev->pdata.mode & SHDMA_DMAOR1)
-		sh_dmae_ctl_stop(1);
+	sh_dmae_ctl_stop(shdev);
 }
 
 static struct platform_driver sh_dmae_driver = {
diff --git a/drivers/dma/shdma.h b/drivers/dma/shdma.h
index 7e227f3c87c4..153609a1e96c 100644
--- a/drivers/dma/shdma.h
+++ b/drivers/dma/shdma.h
@@ -17,23 +17,9 @@
 #include <linux/interrupt.h>
 #include <linux/list.h>
 
-#define SH_DMA_TCR_MAX 0x00FFFFFF	/* 16MB */
-
-struct sh_dmae_regs {
-	u32 sar; /* SAR / source address */
-	u32 dar; /* DAR / destination address */
-	u32 tcr; /* TCR / transfer count */
-};
+#include <asm/dmaengine.h>
 
-struct sh_desc {
-	struct sh_dmae_regs hw;
-	struct list_head node;
-	struct dma_async_tx_descriptor async_tx;
-	enum dma_data_direction direction;
-	dma_cookie_t cookie;
-	int chunks;
-	int mark;
-};
+#define SH_DMA_TCR_MAX 0x00FFFFFF	/* 16MB */
 
 struct device;
 
@@ -47,14 +33,18 @@ struct sh_dmae_chan {
 	struct tasklet_struct tasklet;	/* Tasklet */
 	int descs_allocated;		/* desc count */
 	int xmit_shift;			/* log_2(bytes_per_xfer) */
+	int irq;
 	int id;				/* Raw id of this channel */
+	u32 __iomem *base;
 	char dev_id[16];		/* unique name per DMAC of channel */
 };
 
 struct sh_dmae_device {
 	struct dma_device common;
-	struct sh_dmae_chan *chan[MAX_DMA_CHANNELS];
-	struct sh_dmae_pdata pdata;
+	struct sh_dmae_chan *chan[SH_DMAC_MAX_CHANNELS];
+	struct sh_dmae_pdata *pdata;
+	u32 __iomem *chan_reg;
+	u16 __iomem *dmars;
 };
 
 #define to_sh_chan(chan) container_of(chan, struct sh_dmae_chan, common)
diff --git a/drivers/serial/Kconfig b/drivers/serial/Kconfig
index 746e07033dce..d6ff73395623 100644
--- a/drivers/serial/Kconfig
+++ b/drivers/serial/Kconfig
@@ -1009,6 +1009,10 @@ config SERIAL_SH_SCI_CONSOLE
 	depends on SERIAL_SH_SCI=y
 	select SERIAL_CORE_CONSOLE
 
+config SERIAL_SH_SCI_DMA
+	bool "DMA support"
+	depends on SERIAL_SH_SCI && SH_DMAE && EXPERIMENTAL
+
 config SERIAL_PNX8XXX
 	bool "Enable PNX8XXX SoCs' UART Support"
 	depends on MIPS && (SOC_PNX8550 || SOC_PNX833X)
diff --git a/drivers/serial/sh-sci.c b/drivers/serial/sh-sci.c
index 42f3333c4ad0..980f39449ee5 100644
--- a/drivers/serial/sh-sci.c
+++ b/drivers/serial/sh-sci.c
@@ -48,6 +48,9 @@
 #include <linux/ctype.h>
 #include <linux/err.h>
 #include <linux/list.h>
+#include <linux/dmaengine.h>
+#include <linux/scatterlist.h>
+#include <linux/timer.h>
 
 #ifdef CONFIG_SUPERH
 #include <asm/sh_bios.h>
@@ -84,6 +87,27 @@ struct sci_port {
 	struct clk		*dclk;
 
 	struct list_head	node;
+	struct dma_chan			*chan_tx;
+	struct dma_chan			*chan_rx;
+#ifdef CONFIG_SERIAL_SH_SCI_DMA
+	struct device			*dma_dev;
+	enum sh_dmae_slave_chan_id	slave_tx;
+	enum sh_dmae_slave_chan_id	slave_rx;
+	struct dma_async_tx_descriptor	*desc_tx;
+	struct dma_async_tx_descriptor	*desc_rx[2];
+	dma_cookie_t			cookie_tx;
+	dma_cookie_t			cookie_rx[2];
+	dma_cookie_t			active_rx;
+	struct scatterlist		sg_tx;
+	unsigned int			sg_len_tx;
+	struct scatterlist		sg_rx[2];
+	size_t				buf_len_rx;
+	struct sh_dmae_slave		param_tx;
+	struct sh_dmae_slave		param_rx;
+	struct work_struct		work_tx;
+	struct work_struct		work_rx;
+	struct timer_list		rx_timer;
+#endif
 };
 
 struct sh_sci_priv {
@@ -269,29 +293,44 @@ static inline void sci_init_pins(struct uart_port *port, unsigned int cflag)
     defined(CONFIG_CPU_SUBTYPE_SH7780) || \
     defined(CONFIG_CPU_SUBTYPE_SH7785) || \
     defined(CONFIG_CPU_SUBTYPE_SH7786)
-static inline int scif_txroom(struct uart_port *port)
+static int scif_txfill(struct uart_port *port)
 {
-	return SCIF_TXROOM_MAX - (sci_in(port, SCTFDR) & 0xff);
+	return sci_in(port, SCTFDR) & 0xff;
 }
 
-static inline int scif_rxroom(struct uart_port *port)
+static int scif_txroom(struct uart_port *port)
+{
+	return SCIF_TXROOM_MAX - scif_txfill(port);
+}
+
+static int scif_rxfill(struct uart_port *port)
 {
 	return sci_in(port, SCRFDR) & 0xff;
 }
 #elif defined(CONFIG_CPU_SUBTYPE_SH7763)
-static inline int scif_txroom(struct uart_port *port)
+static int scif_txfill(struct uart_port *port)
 {
-	if ((port->mapbase == 0xffe00000) ||
-	    (port->mapbase == 0xffe08000)) {
+	if (port->mapbase == 0xffe00000 ||
+	    port->mapbase == 0xffe08000)
 		/* SCIF0/1*/
-		return SCIF_TXROOM_MAX - (sci_in(port, SCTFDR) & 0xff);
-	} else {
+		return sci_in(port, SCTFDR) & 0xff;
+	else
 		/* SCIF2 */
-		return SCIF2_TXROOM_MAX - (sci_in(port, SCFDR) >> 8);
-	}
+		return sci_in(port, SCFDR) >> 8;
 }
 
-static inline int scif_rxroom(struct uart_port *port)
+static int scif_txroom(struct uart_port *port)
+{
+	if (port->mapbase == 0xffe00000 ||
+	    port->mapbase == 0xffe08000)
+		/* SCIF0/1*/
+		return SCIF_TXROOM_MAX - scif_txfill(port);
+	else
+		/* SCIF2 */
+		return SCIF2_TXROOM_MAX - scif_txfill(port);
+}
+
+static int scif_rxfill(struct uart_port *port)
 {
 	if ((port->mapbase == 0xffe00000) ||
 	    (port->mapbase == 0xffe08000)) {
@@ -303,23 +342,33 @@ static inline int scif_rxroom(struct uart_port *port)
 	}
 }
 #else
-static inline int scif_txroom(struct uart_port *port)
+static int scif_txfill(struct uart_port *port)
+{
+	return sci_in(port, SCFDR) >> 8;
+}
+
+static int scif_txroom(struct uart_port *port)
 {
-	return SCIF_TXROOM_MAX - (sci_in(port, SCFDR) >> 8);
+	return SCIF_TXROOM_MAX - scif_txfill(port);
 }
 
-static inline int scif_rxroom(struct uart_port *port)
+static int scif_rxfill(struct uart_port *port)
 {
 	return sci_in(port, SCFDR) & SCIF_RFDC_MASK;
 }
 #endif
 
-static inline int sci_txroom(struct uart_port *port)
+static int sci_txfill(struct uart_port *port)
 {
-	return (sci_in(port, SCxSR) & SCI_TDRE) != 0;
+	return !(sci_in(port, SCxSR) & SCI_TDRE);
 }
 
-static inline int sci_rxroom(struct uart_port *port)
+static int sci_txroom(struct uart_port *port)
+{
+	return !sci_txfill(port);
+}
+
+static int sci_rxfill(struct uart_port *port)
 {
 	return (sci_in(port, SCxSR) & SCxSR_RDxF(port)) != 0;
 }
@@ -406,9 +455,9 @@ static inline void sci_receive_chars(struct uart_port *port)
 
 	while (1) {
 		if (port->type == PORT_SCI)
-			count = sci_rxroom(port);
+			count = sci_rxfill(port);
 		else
-			count = scif_rxroom(port);
+			count = scif_rxfill(port);
 
 		/* Don't copy more bytes than there is room for in the buffer */
 		count = tty_buffer_request_room(tty, count);
@@ -453,10 +502,10 @@ static inline void sci_receive_chars(struct uart_port *port)
 				}
 
 				/* Store data and status */
-				if (status&SCxSR_FER(port)) {
+				if (status & SCxSR_FER(port)) {
 					flag = TTY_FRAME;
 					dev_notice(port->dev, "frame error\n");
-				} else if (status&SCxSR_PER(port)) {
+				} else if (status & SCxSR_PER(port)) {
 					flag = TTY_PARITY;
 					dev_notice(port->dev, "parity error\n");
 				} else
@@ -618,13 +667,39 @@ static inline int sci_handle_breaks(struct uart_port *port)
 	return copied;
 }
 
-static irqreturn_t sci_rx_interrupt(int irq, void *port)
+static irqreturn_t sci_rx_interrupt(int irq, void *ptr)
 {
+#ifdef CONFIG_SERIAL_SH_SCI_DMA
+	struct uart_port *port = ptr;
+	struct sci_port *s = to_sci_port(port);
+
+	if (s->chan_rx) {
+		unsigned long tout;
+		u16 scr = sci_in(port, SCSCR);
+		u16 ssr = sci_in(port, SCxSR);
+
+		/* Disable future Rx interrupts */
+		sci_out(port, SCSCR, scr & ~SCI_CTRL_FLAGS_RIE);
+		/* Clear current interrupt */
+		sci_out(port, SCxSR, ssr & ~(1 | SCxSR_RDxF(port)));
+		/* Calculate delay for 1.5 DMA buffers */
+		tout = (port->timeout - HZ / 50) * s->buf_len_rx * 3 /
+			port->fifosize / 2;
+		dev_dbg(port->dev, "Rx IRQ: setup timeout in %lu ms\n",
+			tout * 1000 / HZ);
+		if (tout < 2)
+			tout = 2;
+		mod_timer(&s->rx_timer, jiffies + tout);
+
+		return IRQ_HANDLED;
+	}
+#endif
+
 	/* I think sci_receive_chars has to be called irrespective
 	 * of whether the I_IXOFF is set, otherwise, how is the interrupt
 	 * to be disabled?
 	 */
-	sci_receive_chars(port);
+	sci_receive_chars(ptr);
 
 	return IRQ_HANDLED;
 }
@@ -680,6 +755,7 @@ static irqreturn_t sci_mpxed_interrupt(int irq, void *ptr)
 {
 	unsigned short ssr_status, scr_status, err_enabled;
 	struct uart_port *port = ptr;
+	struct sci_port *s = to_sci_port(port);
 	irqreturn_t ret = IRQ_NONE;
 
 	ssr_status = sci_in(port, SCxSR);
@@ -687,10 +763,15 @@ static irqreturn_t sci_mpxed_interrupt(int irq, void *ptr)
 	err_enabled = scr_status & (SCI_CTRL_FLAGS_REIE | SCI_CTRL_FLAGS_RIE);
 
 	/* Tx Interrupt */
-	if ((ssr_status & SCxSR_TDxE(port)) && (scr_status & SCI_CTRL_FLAGS_TIE))
+	if ((ssr_status & SCxSR_TDxE(port)) && (scr_status & SCI_CTRL_FLAGS_TIE) &&
+	    !s->chan_tx)
 		ret = sci_tx_interrupt(irq, ptr);
-	/* Rx Interrupt */
-	if ((ssr_status & SCxSR_RDxF(port)) && (scr_status & SCI_CTRL_FLAGS_RIE))
+	/*
+	 * Rx Interrupt: if we're using DMA, the DMA controller clears RDF /
+	 * DR flags
+	 */
+	if (((ssr_status & SCxSR_RDxF(port)) || s->chan_rx) &&
+	    (scr_status & SCI_CTRL_FLAGS_RIE))
 		ret = sci_rx_interrupt(irq, ptr);
 	/* Error Interrupt */
 	if ((ssr_status & SCxSR_ERRORS(port)) && err_enabled)
@@ -699,6 +780,10 @@ static irqreturn_t sci_mpxed_interrupt(int irq, void *ptr)
 	if ((ssr_status & SCxSR_BRK(port)) && err_enabled)
 		ret = sci_br_interrupt(irq, ptr);
 
+	WARN_ONCE(ret == IRQ_NONE,
+		  "%s: %d IRQ %d, status %x, control %x\n", __func__,
+		  irq, port->line, ssr_status, scr_status);
+
 	return ret;
 }
 
@@ -800,7 +885,9 @@ static void sci_free_irq(struct sci_port *port)
 static unsigned int sci_tx_empty(struct uart_port *port)
 {
 	unsigned short status = sci_in(port, SCxSR);
-	return status & SCxSR_TEND(port) ? TIOCSER_TEMT : 0;
+	unsigned short in_tx_fifo = scif_txfill(port);
+
+	return (status & SCxSR_TEND(port)) && !in_tx_fifo ? TIOCSER_TEMT : 0;
 }
 
 static void sci_set_mctrl(struct uart_port *port, unsigned int mctrl)
@@ -812,16 +899,297 @@ static void sci_set_mctrl(struct uart_port *port, unsigned int mctrl)
 
 static unsigned int sci_get_mctrl(struct uart_port *port)
 {
-	/* This routine is used for geting signals of: DTR, DCD, DSR, RI,
+	/* This routine is used for getting signals of: DTR, DCD, DSR, RI,
 	   and CTS/RTS */
 
 	return TIOCM_DTR | TIOCM_RTS | TIOCM_DSR;
 }
 
+#ifdef CONFIG_SERIAL_SH_SCI_DMA
+static void sci_dma_tx_complete(void *arg)
+{
+	struct sci_port *s = arg;
+	struct uart_port *port = &s->port;
+	struct circ_buf *xmit = &port->state->xmit;
+	unsigned long flags;
+
+	dev_dbg(port->dev, "%s(%d)\n", __func__, port->line);
+
+	spin_lock_irqsave(&port->lock, flags);
+
+	xmit->tail += s->sg_tx.length;
+	xmit->tail &= UART_XMIT_SIZE - 1;
+
+	port->icount.tx += s->sg_tx.length;
+
+	async_tx_ack(s->desc_tx);
+	s->cookie_tx = -EINVAL;
+	s->desc_tx = NULL;
+
+	spin_unlock_irqrestore(&port->lock, flags);
+
+	if (uart_circ_chars_pending(xmit) < WAKEUP_CHARS)
+		uart_write_wakeup(port);
+
+	if (uart_circ_chars_pending(xmit))
+		schedule_work(&s->work_tx);
+}
+
+/* Locking: called with port lock held */
+static int sci_dma_rx_push(struct sci_port *s, struct tty_struct *tty,
+			   size_t count)
+{
+	struct uart_port *port = &s->port;
+	int i, active, room;
+
+	room = tty_buffer_request_room(tty, count);
+
+	if (s->active_rx == s->cookie_rx[0]) {
+		active = 0;
+	} else if (s->active_rx == s->cookie_rx[1]) {
+		active = 1;
+	} else {
+		dev_err(port->dev, "cookie %d not found!\n", s->active_rx);
+		return 0;
+	}
+
+	if (room < count)
+		dev_warn(port->dev, "Rx overrun: dropping %u bytes\n",
+			 count - room);
+	if (!room)
+		return room;
+
+	for (i = 0; i < room; i++)
+		tty_insert_flip_char(tty, ((u8 *)sg_virt(&s->sg_rx[active]))[i],
+				     TTY_NORMAL);
+
+	port->icount.rx += room;
+
+	return room;
+}
+
+static void sci_dma_rx_complete(void *arg)
+{
+	struct sci_port *s = arg;
+	struct uart_port *port = &s->port;
+	struct tty_struct *tty = port->state->port.tty;
+	unsigned long flags;
+	int count;
+
+	dev_dbg(port->dev, "%s(%d)\n", __func__, port->line);
+
+	spin_lock_irqsave(&port->lock, flags);
+
+	count = sci_dma_rx_push(s, tty, s->buf_len_rx);
+
+	mod_timer(&s->rx_timer, jiffies + msecs_to_jiffies(5));
+
+	spin_unlock_irqrestore(&port->lock, flags);
+
+	if (count)
+		tty_flip_buffer_push(tty);
+
+	schedule_work(&s->work_rx);
+}
+
+static void sci_start_rx(struct uart_port *port);
+static void sci_start_tx(struct uart_port *port);
+
+static void sci_rx_dma_release(struct sci_port *s, bool enable_pio)
+{
+	struct dma_chan *chan = s->chan_rx;
+	struct uart_port *port = &s->port;
+
+	s->chan_rx = NULL;
+	s->cookie_rx[0] = s->cookie_rx[1] = -EINVAL;
+	dma_release_channel(chan);
+	dma_free_coherent(port->dev, s->buf_len_rx * 2,
+			  sg_virt(&s->sg_rx[0]), sg_dma_address(&s->sg_rx[0]));
+	if (enable_pio)
+		sci_start_rx(port);
+}
+
+static void sci_tx_dma_release(struct sci_port *s, bool enable_pio)
+{
+	struct dma_chan *chan = s->chan_tx;
+	struct uart_port *port = &s->port;
+
+	s->chan_tx = NULL;
+	s->cookie_tx = -EINVAL;
+	dma_release_channel(chan);
+	if (enable_pio)
+		sci_start_tx(port);
+}
+
+static void sci_submit_rx(struct sci_port *s)
+{
+	struct dma_chan *chan = s->chan_rx;
+	int i;
+
+	for (i = 0; i < 2; i++) {
+		struct scatterlist *sg = &s->sg_rx[i];
+		struct dma_async_tx_descriptor *desc;
+
+		desc = chan->device->device_prep_slave_sg(chan,
+			sg, 1, DMA_FROM_DEVICE, DMA_PREP_INTERRUPT);
+
+		if (desc) {
+			s->desc_rx[i] = desc;
+			desc->callback = sci_dma_rx_complete;
+			desc->callback_param = s;
+			s->cookie_rx[i] = desc->tx_submit(desc);
+		}
+
+		if (!desc || s->cookie_rx[i] < 0) {
+			if (i) {
+				async_tx_ack(s->desc_rx[0]);
+				s->cookie_rx[0] = -EINVAL;
+			}
+			if (desc) {
+				async_tx_ack(desc);
+				s->cookie_rx[i] = -EINVAL;
+			}
+			dev_warn(s->port.dev,
+				 "failed to re-start DMA, using PIO\n");
+			sci_rx_dma_release(s, true);
+			return;
+		}
+	}
+
+	s->active_rx = s->cookie_rx[0];
+
+	dma_async_issue_pending(chan);
+}
+
+static void work_fn_rx(struct work_struct *work)
+{
+	struct sci_port *s = container_of(work, struct sci_port, work_rx);
+	struct uart_port *port = &s->port;
+	struct dma_async_tx_descriptor *desc;
+	int new;
+
+	if (s->active_rx == s->cookie_rx[0]) {
+		new = 0;
+	} else if (s->active_rx == s->cookie_rx[1]) {
+		new = 1;
+	} else {
+		dev_err(port->dev, "cookie %d not found!\n", s->active_rx);
+		return;
+	}
+	desc = s->desc_rx[new];
+
+	if (dma_async_is_tx_complete(s->chan_rx, s->active_rx, NULL, NULL) !=
+	    DMA_SUCCESS) {
+		/* Handle incomplete DMA receive */
+		struct tty_struct *tty = port->state->port.tty;
+		struct dma_chan *chan = s->chan_rx;
+		struct sh_desc *sh_desc = container_of(desc, struct sh_desc,
+						       async_tx);
+		unsigned long flags;
+		int count;
+
+		chan->device->device_terminate_all(chan);
+		dev_dbg(port->dev, "Read %u bytes with cookie %d\n",
+			sh_desc->partial, sh_desc->cookie);
+
+		spin_lock_irqsave(&port->lock, flags);
+		count = sci_dma_rx_push(s, tty, sh_desc->partial);
+		spin_unlock_irqrestore(&port->lock, flags);
+
+		if (count)
+			tty_flip_buffer_push(tty);
+
+		sci_submit_rx(s);
+
+		return;
+	}
+
+	s->cookie_rx[new] = desc->tx_submit(desc);
+	if (s->cookie_rx[new] < 0) {
+		dev_warn(port->dev, "Failed submitting Rx DMA descriptor\n");
+		sci_rx_dma_release(s, true);
+		return;
+	}
+
+	dev_dbg(port->dev, "%s: cookie %d #%d\n", __func__,
+		s->cookie_rx[new], new);
+
+	s->active_rx = s->cookie_rx[!new];
+}
+
+static void work_fn_tx(struct work_struct *work)
+{
+	struct sci_port *s = container_of(work, struct sci_port, work_tx);
+	struct dma_async_tx_descriptor *desc;
+	struct dma_chan *chan = s->chan_tx;
+	struct uart_port *port = &s->port;
+	struct circ_buf *xmit = &port->state->xmit;
+	struct scatterlist *sg = &s->sg_tx;
+
+	/*
+	 * DMA is idle now.
+	 * Port xmit buffer is already mapped, and it is one page... Just adjust
+	 * offsets and lengths. Since it is a circular buffer, we have to
+	 * transmit till the end, and then the rest. Take the port lock to get a
+	 * consistent xmit buffer state.
+	 */
+	spin_lock_irq(&port->lock);
+	sg->offset = xmit->tail & (UART_XMIT_SIZE - 1);
+	sg->dma_address = (sg_dma_address(sg) & ~(UART_XMIT_SIZE - 1)) +
+		sg->offset;
+	sg->length = min((int)CIRC_CNT(xmit->head, xmit->tail, UART_XMIT_SIZE),
+		CIRC_CNT_TO_END(xmit->head, xmit->tail, UART_XMIT_SIZE));
+	sg->dma_length = sg->length;
+	spin_unlock_irq(&port->lock);
+
+	BUG_ON(!sg->length);
+
+	desc = chan->device->device_prep_slave_sg(chan,
+			sg, s->sg_len_tx, DMA_TO_DEVICE,
+			DMA_PREP_INTERRUPT | DMA_CTRL_ACK);
+	if (!desc) {
+		/* switch to PIO */
+		sci_tx_dma_release(s, true);
+		return;
+	}
+
+	dma_sync_sg_for_device(port->dev, sg, 1, DMA_TO_DEVICE);
+
+	spin_lock_irq(&port->lock);
+	s->desc_tx = desc;
+	desc->callback = sci_dma_tx_complete;
+	desc->callback_param = s;
+	spin_unlock_irq(&port->lock);
+	s->cookie_tx = desc->tx_submit(desc);
+	if (s->cookie_tx < 0) {
+		dev_warn(port->dev, "Failed submitting Tx DMA descriptor\n");
+		/* switch to PIO */
+		sci_tx_dma_release(s, true);
+		return;
+	}
+
+	dev_dbg(port->dev, "%s: %p: %d...%d, cookie %d\n", __func__,
+		xmit->buf, xmit->tail, xmit->head, s->cookie_tx);
+
+	dma_async_issue_pending(chan);
+}
+#endif
+
 static void sci_start_tx(struct uart_port *port)
 {
 	unsigned short ctrl;
 
+#ifdef CONFIG_SERIAL_SH_SCI_DMA
+	struct sci_port *s = to_sci_port(port);
+
+	if (s->chan_tx) {
+		if (!uart_circ_empty(&s->port.state->xmit) && s->cookie_tx < 0)
+			schedule_work(&s->work_tx);
+
+		return;
+	}
+#endif
+
 	/* Set TIE (Transmit Interrupt Enable) bit in SCSCR */
 	ctrl = sci_in(port, SCSCR);
 	ctrl |= SCI_CTRL_FLAGS_TIE;
@@ -838,13 +1206,12 @@ static void sci_stop_tx(struct uart_port *port)
 	sci_out(port, SCSCR, ctrl);
 }
 
-static void sci_start_rx(struct uart_port *port, unsigned int tty_start)
+static void sci_start_rx(struct uart_port *port)
 {
-	unsigned short ctrl;
+	unsigned short ctrl = SCI_CTRL_FLAGS_RIE | SCI_CTRL_FLAGS_REIE;
 
 	/* Set RIE (Receive Interrupt Enable) bit in SCSCR */
-	ctrl = sci_in(port, SCSCR);
-	ctrl |= SCI_CTRL_FLAGS_RIE | SCI_CTRL_FLAGS_REIE;
+	ctrl |= sci_in(port, SCSCR);
 	sci_out(port, SCSCR, ctrl);
 }
 
@@ -868,16 +1235,154 @@ static void sci_break_ctl(struct uart_port *port, int break_state)
 	/* Nothing here yet .. */
 }
 
+#ifdef CONFIG_SERIAL_SH_SCI_DMA
+static bool filter(struct dma_chan *chan, void *slave)
+{
+	struct sh_dmae_slave *param = slave;
+
+	dev_dbg(chan->device->dev, "%s: slave ID %d\n", __func__,
+		param->slave_id);
+
+	if (param->dma_dev == chan->device->dev) {
+		chan->private = param;
+		return true;
+	} else {
+		return false;
+	}
+}
+
+static void rx_timer_fn(unsigned long arg)
+{
+	struct sci_port *s = (struct sci_port *)arg;
+	struct uart_port *port = &s->port;
+
+	u16 scr = sci_in(port, SCSCR);
+	sci_out(port, SCSCR, scr | SCI_CTRL_FLAGS_RIE);
+	dev_dbg(port->dev, "DMA Rx timed out\n");
+	schedule_work(&s->work_rx);
+}
+
+static void sci_request_dma(struct uart_port *port)
+{
+	struct sci_port *s = to_sci_port(port);
+	struct sh_dmae_slave *param;
+	struct dma_chan *chan;
+	dma_cap_mask_t mask;
+	int nent;
+
+	dev_dbg(port->dev, "%s: port %d DMA %p\n", __func__,
+		port->line, s->dma_dev);
+
+	if (!s->dma_dev)
+		return;
+
+	dma_cap_zero(mask);
+	dma_cap_set(DMA_SLAVE, mask);
+
+	param = &s->param_tx;
+
+	/* Slave ID, e.g., SHDMA_SLAVE_SCIF0_TX */
+	param->slave_id = s->slave_tx;
+	param->dma_dev = s->dma_dev;
+
+	s->cookie_tx = -EINVAL;
+	chan = dma_request_channel(mask, filter, param);
+	dev_dbg(port->dev, "%s: TX: got channel %p\n", __func__, chan);
+	if (chan) {
+		s->chan_tx = chan;
+		sg_init_table(&s->sg_tx, 1);
+		/* UART circular tx buffer is an aligned page. */
+		BUG_ON((int)port->state->xmit.buf & ~PAGE_MASK);
+		sg_set_page(&s->sg_tx, virt_to_page(port->state->xmit.buf),
+			    UART_XMIT_SIZE, (int)port->state->xmit.buf & ~PAGE_MASK);
+		nent = dma_map_sg(port->dev, &s->sg_tx, 1, DMA_TO_DEVICE);
+		if (!nent)
+			sci_tx_dma_release(s, false);
+		else
+			dev_dbg(port->dev, "%s: mapped %d@%p to %x\n", __func__,
+				sg_dma_len(&s->sg_tx),
+				port->state->xmit.buf, sg_dma_address(&s->sg_tx));
+
+		s->sg_len_tx = nent;
+
+		INIT_WORK(&s->work_tx, work_fn_tx);
+	}
+
+	param = &s->param_rx;
+
+	/* Slave ID, e.g., SHDMA_SLAVE_SCIF0_RX */
+	param->slave_id = s->slave_rx;
+	param->dma_dev = s->dma_dev;
+
+	chan = dma_request_channel(mask, filter, param);
+	dev_dbg(port->dev, "%s: RX: got channel %p\n", __func__, chan);
+	if (chan) {
+		dma_addr_t dma[2];
+		void *buf[2];
+		int i;
+
+		s->chan_rx = chan;
+
+		s->buf_len_rx = 2 * max(16, (int)port->fifosize);
+		buf[0] = dma_alloc_coherent(port->dev, s->buf_len_rx * 2,
+					    &dma[0], GFP_KERNEL);
+
+		if (!buf[0]) {
+			dev_warn(port->dev,
+				 "failed to allocate dma buffer, using PIO\n");
+			sci_rx_dma_release(s, true);
+			return;
+		}
+
+		buf[1] = buf[0] + s->buf_len_rx;
+		dma[1] = dma[0] + s->buf_len_rx;
+
+		for (i = 0; i < 2; i++) {
+			struct scatterlist *sg = &s->sg_rx[i];
+
+			sg_init_table(sg, 1);
+			sg_set_page(sg, virt_to_page(buf[i]), s->buf_len_rx,
+				    (int)buf[i] & ~PAGE_MASK);
+			sg->dma_address = dma[i];
+			sg->dma_length = sg->length;
+		}
+
+		INIT_WORK(&s->work_rx, work_fn_rx);
+		setup_timer(&s->rx_timer, rx_timer_fn, (unsigned long)s);
+
+		sci_submit_rx(s);
+	}
+}
+
+static void sci_free_dma(struct uart_port *port)
+{
+	struct sci_port *s = to_sci_port(port);
+
+	if (!s->dma_dev)
+		return;
+
+	if (s->chan_tx)
+		sci_tx_dma_release(s, false);
+	if (s->chan_rx)
+		sci_rx_dma_release(s, false);
+}
+#endif
+
 static int sci_startup(struct uart_port *port)
 {
 	struct sci_port *s = to_sci_port(port);
 
+	dev_dbg(port->dev, "%s(%d)\n", __func__, port->line);
+
 	if (s->enable)
 		s->enable(port);
 
 	sci_request_irq(s);
+#ifdef CONFIG_SERIAL_SH_SCI_DMA
+	sci_request_dma(port);
+#endif
 	sci_start_tx(port);
-	sci_start_rx(port, 1);
+	sci_start_rx(port);
 
 	return 0;
 }
@@ -886,8 +1391,13 @@ static void sci_shutdown(struct uart_port *port)
 {
 	struct sci_port *s = to_sci_port(port);
 
+	dev_dbg(port->dev, "%s(%d)\n", __func__, port->line);
+
 	sci_stop_rx(port);
 	sci_stop_tx(port);
+#ifdef CONFIG_SERIAL_SH_SCI_DMA
+	sci_free_dma(port);
+#endif
 	sci_free_irq(s);
 
 	if (s->disable)
@@ -937,6 +1447,9 @@ static void sci_set_termios(struct uart_port *port, struct ktermios *termios,
 
 	sci_out(port, SCSMR, smr_val);
 
+	dev_dbg(port->dev, "%s: SMR %x, t %x, SCSCR %x\n", __func__, smr_val, t,
+		SCSCR_INIT(port));
+
 	if (t > 0) {
 		if (t >= 256) {
 			sci_out(port, SCSMR, (sci_in(port, SCSMR) & ~3) | 1);
@@ -954,7 +1467,7 @@ static void sci_set_termios(struct uart_port *port, struct ktermios *termios,
 	sci_out(port, SCSCR, SCSCR_INIT(port));
 
 	if ((termios->c_cflag & CREAD) != 0)
-		sci_start_rx(port, 0);
+		sci_start_rx(port);
 }
 
 static const char *sci_type(struct uart_port *port)
@@ -1049,19 +1562,21 @@ static void __devinit sci_init_single(struct platform_device *dev,
 				      unsigned int index,
 				      struct plat_sci_port *p)
 {
-	sci_port->port.ops	= &sci_uart_ops;
-	sci_port->port.iotype	= UPIO_MEM;
-	sci_port->port.line	= index;
+	struct uart_port *port = &sci_port->port;
+
+	port->ops	= &sci_uart_ops;
+	port->iotype	= UPIO_MEM;
+	port->line	= index;
 
 	switch (p->type) {
 	case PORT_SCIFA:
-		sci_port->port.fifosize = 64;
+		port->fifosize = 64;
 		break;
 	case PORT_SCIF:
-		sci_port->port.fifosize = 16;
+		port->fifosize = 16;
 		break;
 	default:
-		sci_port->port.fifosize = 1;
+		port->fifosize = 1;
 		break;
 	}
 
@@ -1070,19 +1585,28 @@ static void __devinit sci_init_single(struct platform_device *dev,
 		sci_port->dclk = clk_get(&dev->dev, "peripheral_clk");
 		sci_port->enable = sci_clk_enable;
 		sci_port->disable = sci_clk_disable;
-		sci_port->port.dev = &dev->dev;
+		port->dev = &dev->dev;
 	}
 
 	sci_port->break_timer.data = (unsigned long)sci_port;
 	sci_port->break_timer.function = sci_break_timer;
 	init_timer(&sci_port->break_timer);
 
-	sci_port->port.mapbase	= p->mapbase;
-	sci_port->port.membase	= p->membase;
+	port->mapbase	= p->mapbase;
+	port->membase	= p->membase;
 
-	sci_port->port.irq	= p->irqs[SCIx_TXI_IRQ];
-	sci_port->port.flags	= p->flags;
-	sci_port->type		= sci_port->port.type = p->type;
+	port->irq	= p->irqs[SCIx_TXI_IRQ];
+	port->flags	= p->flags;
+	sci_port->type	= port->type = p->type;
+
+#ifdef CONFIG_SERIAL_SH_SCI_DMA
+	sci_port->dma_dev	= p->dma_dev;
+	sci_port->slave_tx	= p->dma_slave_tx;
+	sci_port->slave_rx	= p->dma_slave_rx;
+
+	dev_dbg(port->dev, "%s: DMA device %p, tx %d, rx %d\n", __func__,
+		p->dma_dev, p->dma_slave_tx, p->dma_slave_rx);
+#endif
 
 	memcpy(&sci_port->irqs, &p->irqs, sizeof(p->irqs));
 }
diff --git a/include/linux/serial_sci.h b/include/linux/serial_sci.h
index 1c297ddc9d5a..1b177d29a7f0 100644
--- a/include/linux/serial_sci.h
+++ b/include/linux/serial_sci.h
@@ -2,6 +2,7 @@
 #define __LINUX_SERIAL_SCI_H
 
 #include <linux/serial_core.h>
+#include <asm/dmaengine.h>
 
 /*
  * Generic header for SuperH SCI(F) (used by sh/sh64/h8300 and related parts)
@@ -16,6 +17,8 @@ enum {
 	SCIx_NR_IRQS,
 };
 
+struct device;
+
 /*
  * Platform device specific platform_data struct
  */
@@ -26,6 +29,9 @@ struct plat_sci_port {
 	unsigned int	type;			/* SCI / SCIF / IRDA */
 	upf_t		flags;			/* UPF_* flags */
 	char		*clk;			/* clock string */
+	struct device	*dma_dev;
+	enum sh_dmae_slave_chan_id dma_slave_tx;
+	enum sh_dmae_slave_chan_id dma_slave_rx;
 };
 
 #endif /* __LINUX_SERIAL_SCI_H */
diff --git a/sound/soc/sh/siu.h b/sound/soc/sh/siu.h
index 9cc04ab2bce7..c0bfab8fed3d 100644
--- a/sound/soc/sh/siu.h
+++ b/sound/soc/sh/siu.h
@@ -72,7 +72,7 @@ struct siu_firmware {
 #include <linux/interrupt.h>
 #include <linux/io.h>
 
-#include <asm/dma-sh.h>
+#include <asm/dmaengine.h>
 
 #include <sound/core.h>
 #include <sound/pcm.h>
diff --git a/sound/soc/sh/siu_pcm.c b/sound/soc/sh/siu_pcm.c
index c5efc30f0136..ba7f8d05d977 100644
--- a/sound/soc/sh/siu_pcm.c
+++ b/sound/soc/sh/siu_pcm.c
@@ -32,7 +32,7 @@
 #include <sound/pcm_params.h>
 #include <sound/soc-dai.h>
 
-#include <asm/dma-sh.h>
+#include <asm/dmaengine.h>
 #include <asm/siu.h>
 
 #include "siu.h"